HOUSE
v.
BELL, WARDEN

*521 Stephen Michael Kissinger argued the cause for petitioner. With him on the briefs were George H. Kendall, Theodore M. Shaw, and Jacqueline A. Berrien.

Jennifer L. Smith, Associate Deputy Attorney General of Tennessee, argued the cause for respondent. With her on the brief were Paul G. Summers, Attorney General, Michael E. Moore, Solicitor General, Gordon W. Smith, Associate Solicitor General, and Alice B. Lustre.^[*]

JUSTICE KENNEDY delivered the opinion of the Court.

Some 20 years ago in rural Tennessee, Carolyn Muncey was murdered. A jury convicted petitioner Paul Gregory House of the crime and sentenced him to death, but new revelations cast doubt on the jury's verdict. House, protesting his innocence, seeks access to federal court to pursue *522 habeas corpus relief based on constitutional claims that are procedurally barred under state law. Out of respect for the finality of state-court judgments federal habeas courts, as a general rule, are closed to claims that state courts would consider defaulted. In certain exceptional cases involving a compelling claim of actual innocence, however, the state procedural default rule is not a bar to a federal habeas corpus petition. See Schlup v. Delo, 513 U.S. 298, 319-322 (1995). After careful review of the full record, we conclude that House has made the stringent showing required by this exception; and we hold that his federal habeas action may proceed.

I

We begin with the facts surrounding Mrs. Muncey's disappearance, the discovery of her body, and House's arrest. Around 3 p.m. on Sunday, July 14, 1985, two local residents found her body concealed amid brush and tree branches on an embankment roughly 100 yards up the road from her driveway. Mrs. Muncey had been seen last on the evening before, when, around 8 p.m., she and her two children—Lora Muncey, aged 10, and Matthew Muncey, aged 8—visited their neighbor, Pam Luttrell. According to Luttrell, Mrs. Muncey mentioned her husband, William Hubert Muncey, Jr., known in the community as "Little Hube" and to his family as "Bubbie." As Luttrell recounted Mrs. Muncey's comment, Mr. Muncey "had gone to dig a grave, and he hadn't come back, but that was all right, because [Mrs. Muncey] was going to make him take her fishing the next day," App. 11-12. Mrs. Muncey returned home, and some time later, before 11 p.m. at the latest, Luttrell "heard a car rev its motor as it went down the road," something Mr. Muncey customarily did when he drove by on his way home. Record, Addendum 4, 5 Tr. of Evidence in No. 378 (Crim. Ct. Union Cty., Tenn.), pp. 641-642 (hereinafter Tr.). Luttrell then went to bed.

Around 1 a.m., Lora and Matthew returned to Luttrell's home, this time with their father, Mr. Muncey, who said his *523 wife was missing. Muncey asked Luttrell to watch the children while he searched for his wife. After he left, Luttrell talked with Lora. According to Luttrell:

"[Lora] said she heard a horn blow, she thought she heard a horn blow, and somebody asked if Bubbie was home, and her mama, you know, told them—no. And then she said she didn't know if she went back to sleep or not, but then she heard her mama going down the steps crying and I am not sure if that is when that she told me that she heard her mama say—oh God, no, not me, or if she told me that the next day, but I do know that she said she heard her mother going down the steps crying." App. 14-15.

While Lora was talking, Luttrell recalled, "Matt kept butting in, you know, on us talking, and he said—sister they said daddy had a wreck, they said daddy had a wreck." Id., at 13.

At House's trial, Lora repeated her account of the night's events, this time referring to the "wreck" her brother had mentioned. To assist in understanding Lora's account, it should be noted that Mrs. Muncey's father-in-law—Little Hube's father—was sometimes called "Big Hube." Lora and her brother called him "Paw Paw." We refer to him as Mr. Muncey, Sr. According to Lora, Mr. Muncey, Sr., had a deep voice, as does petitioner House.

Lora testified that after leaving Luttrell's house with her mother, she and her brother "went to bed." Id., at 18. Later, she heard someone, or perhaps two different people, ask for her mother. Lora's account of the events after she went to bed was as follows:

"Q Laura [sic], at some point after you got back home and you went to bed, did anything happen that caused your mother to be upset or did you hear anything?

"A Well, it sounded like PawPaw said—where's daddy at, and she said digging a grave.

*524 "Q Okay. Do you know if it was PawPaw or not, or did it sound like PawPaw?

"A It just sounded like PawPaw.

"Q And your mother told him what?

"A That he was digging a grave.

"Q Had you ever heard that voice before that said that?

"A I don't remember.

"Q After that, at some point later, did you hear anything else that caused your mother to be upset?

"A Well, they said that daddy had a wreck down the road and she started crying—next to the creek.

"Q Your mother started crying. What was it that they said?

"A That daddy had a wreck.

"Q Did they say where?

"A Down there next to the creek." Id., at 18-19.

Lora did not describe hearing any struggle. Some time later, Lora and her brother left the house to look for their mother, but no one answered when they knocked at the Luttrells' home, and another neighbor, Mike Clinton, said he had not seen her. After the children returned home, according to Lora, her father came home and "fixed him a bologna sandwich and he took a bit of it and he says—sissy, where is mommy at, and I said—she ain't been here for a little while." Id., at 20. Lora recalled that Mr. Muncey went outside and, not seeing his wife, returned to take Lora and Matthew to the Luttrells' so that he could look further.

The next afternoon Billy Ray Hensley, the victim's first cousin, heard of Mrs. Muncey's disappearance and went to look for Mr. Muncey. As he approached the Munceys' street, Hensley allegedly "saw Mr. House come out from under a bank, wiping his hands on a black rag." Id., at 32. Just when and where Hensley saw House, and how well he could *525 have observed him, were disputed at House's trial. Hensley admitted on cross-examination that he could not have seen House "walking up or climbing up" the embankment, id., at 39; rather, he saw House, in "[j]ust a glance," id., at 40, "appear out of nowhere," "next to the embankment," id., at 39. On the Munceys' street, opposite the area where Hensley said he saw House, a white Plymouth was parked near a sawmill. Another witness, Billy Hankins, whom the defense called, claimed that around the same time he saw a "boy" walking down the street away from the parked Plymouth and toward the Munceys' home. This witness, however, put the "boy" on the side of the street with the parked car and the Munceys' driveway, not the side with the embankment.

Hensley, after turning onto the Munceys' street, continued down the road and turned into their driveway. "I pulled up in the driveway where I could see up toward Little Hube's house," Hensley testified, "and I seen Little Hube's car wasn't there, and I backed out in the road, and come back [the other way]." Id., at 32. As he traveled up the road, Hensley saw House traveling in the opposite direction in the white Plymouth. House "flagged [Hensley] down" through his windshield, ibid., and the two cars met about 300 feet up the road from the Munceys' driveway. According to Hensley, House said he had heard Mrs. Muncey was missing and was looking for her husband. Id., at 33. Though House had only recently moved to the area, he was acquainted with the Munceys, had attended a dance with them, and had visited their home. He later told law enforcement officials he considered both of the Munceys his friends. According to Hensley, House said he had heard that Mrs. Muncey's husband, who was an alcoholic, was elsewhere "getting drunk." Ibid.

As Hensley drove off, he "got to thinking to [him]self—he's hunting Little Hube, and Little Hube drunk—what would he be doing off that bank ... ." Ibid. His suspicion aroused, Hensley later returned to the Munceys' street with a friend named Jack Adkins. The two checked different spots on the *526 embankment, and though Hensley saw nothing where he looked, Adkins found Mrs. Muncey. Her body lay across from the sawmill near the corner where House's car had been parked, dumped in the woods a short way down the bank leading toward a creek.

Around midnight, Dr. Alex Carabia, a practicing pathologist and county medical examiner, performed an autopsy. Dr. Carabia put the time of death between 9 and 11 p.m. Mrs. Muncey had a black eye, both her hands were bloodstained up to the wrists, and she had bruises on her legs and neck. Dr. Carabia described the bruises as consistent with a "traumatic origin," i. e., a fight or a fall on hard objects. 7 Tr. 985-986. Based on the neck bruises and other injuries, he concluded Mrs. Muncey had been choked, but he ruled this out as the cause of death. The cause of death, in Dr. Carabia's view, was a severe blow to the left forehead that inflicted both a laceration penetrating to the bone and, inside the skull, a severe right-side hemorrhage, likely caused by Mrs. Muncey's brain slamming into the skull opposite the impact. Dr. Carabia described this head injury as consistent either with receiving a blow from a fist or other instrument or with striking some object.

The county sheriff, informed about Hensley's earlier encounter with House, questioned House shortly after the body was found. That evening, House answered further questions during a voluntary interview at the local jail. Special Agent Ray Presnell of the Tennessee Bureau of Investigation (TBI) prepared a statement of House's answers, which House signed. Asked to describe his whereabouts on the previous evening, House claimed—falsely, as it turned out— that he spent the entire evening with his girlfriend, Donna Turner, at her trailer. Asked whether he was wearing the same pants he had worn the night before, House replied— again, falsely—that he was. House was on probation at the time, having recently been released on parole following a sentence of five years to life for aggravated sexual assault in *527 Utah. House had scratches on his arms and hands, and a knuckle on his right ring finger was bruised. He attributed the scratches to Turner's cats and the finger injury to recent construction work tearing down a shed. The next day House gave a similar statement to a different TBI agent, Charles Scott.

In fact House had not been at Turner's home. After initially supporting House's alibi, Turner informed authorities that House left her trailer around 10:30 or 10:45 p.m. to go for a walk. According to Turner's trial testimony, House returned later—she was not sure when—hot and panting, missing his shirt and his shoes. House, Turner testified, told her that while he was walking on the road near her home, a vehicle pulled up beside him, and somebody inside "called him some names and then they told him he didn't belong here anymore." App. 89. House said he tried to ignore the taunts and keep walking, but the vehicle pulled in behind him, and "one of them got out and grabbed him by the shoulder... and [House] swung around with his right hand" and "hit something." Ibid. According to Turner, House said "he took off down the bank and started running and he said that he—he said it seemed forever where he was running. And he said they fired two shots at him while he took off down the bank ... ." Ibid. House claimed the assailants "grabbed ahold of his shirt," which Turner remembered as "a blue tank top, trimmed in yellow," and "they tore it to where it wouldn't stay on him and he said—I just throwed it off when I was running." Id., at 91. Turner, noticing House's bruised knuckle, asked how he hurt it, and House told her "that's where he hit." Id., at 90. Turner testified that she "thought maybe my ex-husband had something to do with it." Ibid.

Although the white Plymouth House drove the next day belonged to Turner, Turner insisted House had not used the car that night. No forensic evidence connected the car to the crime; law enforcement officials inspected a white towel *528 covering the driver seat and concluded it was clean. Turner's trailer was located just under two miles by road, through hilly terrain, from the Muncey residence.

Law enforcement officers also questioned the victim's husband. Though Mrs. Muncey's comments to Luttrell gave no indication she knew this, Mr. Muncey had spent the evening at a weekly dance at a recreation center roughly a mile and a half from his home. In his statement to law enforcement—a statement House's trial counsel claims he never saw— Mr. Muncey admitted leaving the dance early, but said it was only for a brief trip to the package store to buy beer. He also stated that he and his wife had had sexual relations Saturday morning.

Late in the evening on Monday, July 15—two days after the murder—law enforcement officers visited Turner's trailer. With Turner's consent, Agent Scott seized the pants House was wearing the night Mrs. Muncey disappeared. The heavily soiled pants were sitting in a laundry hamper; years later, Agent Scott recalled noticing "reddish brown stains" he "suspected" were blood. Id., at 274-275. Around 4 p.m. the next day, two local law enforcement officers set out for the Federal Bureau of Investigation in Washington, D. C., with House's pants, blood samples from the autopsy, and other evidence packed together in a box. They arrived at 2 a.m. the next morning. On July 17, after initial FBI testing revealed human blood on the pants, House was arrested.

II

The State of Tennessee charged House with capital murder. At House's trial, the State presented testimony by Luttrell, Hensley, Adkins, Lora Muncey, Dr. Carabia, the sheriff, and other law enforcement officials. Through TBI Agents Presnell and Scott, the jury learned of House's false statements. Central to the State's case, however, was what the FBI testing showed—that semen consistent (or so it seemed) with House's was present on Mrs. Muncey's nightgown *529 and panties, and that small bloodstains consistent with Mrs. Muncey's blood but not House's appeared on the jeans belonging to House.

Regarding the semen, FBI Special Agent Paul Bigbee, a serologist, testified that the source was a "secretor," meaning someone who "secrete[s] the ABO blood group substances in other body fluids, such as semen and saliva"—a characteristic shared by 80 percent of the population, including House. Id., at 55. Agent Bigbee further testified that the source of semen on the gown was blood-type A, House's own blood type. As to the semen on the panties, Agent Bigbee found only the H blood-group substance, which A and B blood-type secretors secrete along with substances A and B, and which O-type secretors secrete exclusively. Agent Bigbee explained, however—using science an amicus here sharply disputes, see Brief for Innocence Project, Inc., as Amicus Curiae 24-26—that House's A antigens could have "degraded" into H, App. 57-58. Agent Bigbee thus concluded that both semen deposits could have come from House, though he acknowledged that the H antigen could have come from Mrs. Muncey herself if she was a secretor—something he "was not able to determine," id., at 58—and that, while Mr. Muncey was himself blood-type A (as was his wife), Agent Bigbee was again "not able to determine his secretor status," id., at 57. Agent Bigbee acknowledged on cross-examination that "a saliva sample" would have sufficed to determine whether Mr. Muncey was a secretor; the State did not provide such a sample, though it did provide samples of Mr. Muncey's blood. Id., at 62.

As for the blood, Agent Bigbee explained that "spots of blood" appeared "on the left outside leg, the right bottom cuff, on the left thigh and in the right inside pocket and on the lower pocket on the outside." Id., at 48. Agent Bigbee determined that the blood's source was type A (the type shared by House, the victim, and Mr. Muncey). He also successfully tested for the enzyme phosphoglucomutase and the *530 blood serum haptoglobin, both of which "are found in all humans" and carry "slight chemical differences" that vary genetically and "can be grouped to differentiate between two individuals if those types are different." Id., at 49-50. Based on these chemical traces and on the A blood type, Agent Bigbee determined that only some 6.75 percent of the population carry similar blood, that the blood was "consistent" with Mrs. Muncey's (as determined by testing autopsy samples), and that it was "impossible" that the blood came from House. Id., at 48-52.

A different FBI expert, Special Agent Chester Blythe, testified about fiber analysis performed on Mrs. Muncey's clothes and on House's pants. Although Agent Blythe found blue jean fibers on Mrs. Muncey's nightgown, brassiere, housecoat, and panties, and in fingernail scrapings taken from her body (scrapings that also contained trace, unidentifiable amounts of blood), he acknowledged that, as the prosecutor put it in questioning the witness, "blue jean material is common material," so "this doesn't mean that the fibers that were all over the victim's clothing were necessarily from [House's] pair of blue jeans." 6 Tr. 864-865. On House's pants, though cotton garments both transfer and retain fibers readily, Agent Blythe found neither hair nor fiber consistent with the victim's hair or clothing.

In the defense case House called Hankins, Clinton, and Turner, as well as House's mother, who testified that House had talked to her by telephone around 9:30 p.m. on the night of the murder and that he had not used her car that evening. House also called the victim's brother, Ricky Green, as a witness. Green testified that on July 2, roughly two weeks before the murder, Mrs. Muncey called him and "said her and Little Hube had been into it and she said she was wanting to leave Little Hube, she said she was wanting to get out— out of it, and she was scared." 7 id., at 1088. Green recalled that at Christmastime in 1982 he had seen Mr. Muncey strike Mrs. Muncey after returning home drunk.

*531 As Turner informed the jury, House's shoes were found several months after the crime in a field near her home. Turner delivered them to authorities. Though the jury did not learn of this fact (and House's counsel claims he did not either), the State tested the shoes for blood and found none. House's shirt was not found.

The State's closing argument suggested that on the night of her murder, Mrs. Muncey "was deceived ... . She had been told [her husband] had had an accident." 9 id., at 1226. The prosecutor emphasized the FBI's blood analysis, noting that "after running many, many, many tests," Agent Bigbee

"was able to tell you that the blood on the defendant's blue jeans was not his own blood, could not be his own blood. He told you that the blood on the blue jeans was consistent with every characteristic in every respect of the deceased's, Carolyn Muncey's, and that ninety-three (93%) percent of the white population would not have that blood type... . He can't tell you one hundred (100%) percent for certain that it was her blood. But folks, he can sure give you a pretty good—a pretty good indication." Id., at 1235-1236.

In the State's rebuttal, after defense counsel questioned House's motive "to go over and kill a woman that he barely knew[,] [w]ho was still dressed, still clad in her clothes," id., at 1274, the prosecutor referred obliquely to the semen stains. While explaining that legally "it does not make any difference under God's heaven, what the motive was," App. 106, the prosecutor told the jury, "you may have an idea why he did it," ibid.:

"The evidence at the scene which seemed to suggest that he was subjecting this lady to some kind of indignity, why would you get a lady out of her house, late at night, in her night clothes, under the trick that her husband has had a wreck down by the creek? ... Well, it is because either you don't want her to tell what indignities *532 you have subjected her to, or she is unwilling and fights against you, against being subjected to those indignities. In other words, it is either to keep her from telling what you have done to her, or it is that you are trying to get her to do something that she nor any mother on that road would want to do with Mr. House, under those conditions, and you kill her because of her resistance. That is what the evidence at the scene suggests about motive." Id., at 106-107.

In addition the government suggested the black rag Hensley said he saw in House's hands was in fact the missing blue tank top, retrieved by House from the crime scene. And the prosecution reiterated the importance of the blood. "[D]efense counsel," he said, "does not start out discussing the fact that his client had blood on his jeans on the night that Carolyn Muncey was killed... . He doesn't start with the fact that nothing that the defense has introduced in this case explains what blood is doing on his jeans, all over his jeans, that is scientifically, completely different from his blood." Id., at 104-105. The jury found House guilty of murder in the first degree.

The trial advanced to the sentencing phase. As aggravating factors to support a capital sentence, the State sought to prove: (1) that House had previously been convicted of a felony involving the use or threat of violence; (2) that the homicide was especially heinous, atrocious, or cruel in that it involved torture or depravity of mind; and (3) that the murder was committed while House was committing, attempting to commit, or fleeing from the commission of, rape or kidnaping. See Tenn. Code Ann. §§ 39-2-203(i)(2), (5), (7) (1982) (repealed 1989); cf. §§ 39-13-204(i)(2), (5), (7) (2003). After presenting evidence of House's parole status and aggravated sexual assault conviction, the State rested. As mitigation, the defense offered testimony from House's father and mother, as well as evidence, presented through House's mother, that House attempted suicide after the *533 guilt-phase verdict. Before the attempt House wrote his mother a letter professing his innocence.

In closing the State urged the jury to find all three aggravating factors and impose death. As to the kidnaping or rape factor, the prosecution suggested Mrs. Muncey was "decoy[ed] or entic[ed] ... away from her family, and confin[ed] against her will because you know that as she was being beaten to death." 10 Tr. 1410. "We also think," the prosecutor added, "the proof shows strong evidence of attempted sexual molestation of the victim to accompany the taking away and murdering her." Id., at 1410-1411. Later the prosecutor argued, "I think the proof shows in the record that it is more likely than not that having been through the process before and having been convicted of a crime involving the threat of violence, or violence to another person, aggravated sexual assault, that the defendant cannot benefit from the type of rehabilitation that correction departments can provide." Id., at 1413. The jury unanimously found all three aggravating factors and concluded "there are no mitigating circumstances sufficiently substantial to outweigh the statutory aggravating circumstance or circumstances." Id., at 1454. The jury recommended a death sentence, which the trial judge imposed.

III

The Tennessee Supreme Court affirmed House's conviction and sentence, describing the evidence against House as "circumstantial" but "quite strong." State v. House, 743 S.W.2d 141, 143, 144 (1987). Two months later, in a state trial court, House filed a pro se petition for postconviction relief, arguing he received ineffective assistance of counsel at trial. The court-appointed counsel amended the petition to raise other issues, including a challenge to certain jury instructions. At a hearing before the same judge who conducted the trial, House's counsel offered no proof beyond the trial transcript. The trial court dismissed the petition, deeming House's trial counsel adequate and overruling *534 House's other objections. On appeal House's attorney renewed only the jury-instructions argument. In an unpublished opinion the Tennessee Court of Criminal Appeals affirmed, and both the Tennessee Supreme Court and this Court, House v. Tennessee, 498 U.S. 912 (1990), denied review.

House filed a second postconviction petition in state court reasserting his ineffective-assistance claim and seeking investigative and/or expert assistance. After extensive litigation regarding whether House's claims were procedurally defaulted the Tennessee Supreme Court held that House's claims were barred under a state statute providing that claims not raised in prior postconviction proceedings are presumptively waived, Tenn. Code Ann. § 40-30-112 (1990) (repealed 1995), and that courts may not consider grounds for relief "which the court finds should be excluded because they have been waived or previously determined," § 40-30-111 (repealed 1995). See House v. State, 911 S.W.2d 705 (1995). This Court denied certiorari. House v. Tennessee, 517 U.S. 1193 (1996).

House next sought federal habeas relief, asserting numerous claims of ineffective assistance of counsel and prosecutorial misconduct. The United States District Court for the Eastern District of Tennessee, though deeming House's claims procedurally defaulted and granting summary judgment to the State on the majority of House's claims, held an evidentiary hearing to determine whether House fell within the "actual innocence" exception to procedural default that this Court recognized as to substantive offenses in Schlup and as to death sentences in Sawyer v. Whitley, 505 U.S. 333 (1992). Presenting evidence we describe in greater detail below, House attacked the semen and blood evidence used at his trial and presented other evidence, including a putative confession, suggesting that Mr. Muncey, not House, committed the murder. The District Court nevertheless denied relief, holding that House had neither demonstrated actual *535 innocence of the murder under Schlup nor established that he was ineligible for the death penalty under Sawyer.

The Court of Appeals for the Sixth Circuit granted a certificate of appealability under 28 U.S. C. § 2253(c) as to all claims in the habeas petition. On the merits a divided panel affirmed, but its opinion was withdrawn and the case taken en banc. A divided en banc court certified state-law questions to the Tennessee Supreme Court. House v. Bell, 311 F.3d 767 (CA6 2002). Concluding that House had made a compelling showing of actual innocence, and recognizing that in Herrera v. Collins, 506 U.S. 390 (1993), this Court assumed without deciding that "in a capital case a truly persuasive demonstration of `actual innocence' made after trial would render the execution of a defendant unconstitutional, and warrant federal habeas relief if there were no state avenue open to process such a claim," id., at 417, the six-judge majority certified questions to the State Supreme Court, 311 F.3d, at 777-778. The questions sought "to ascertain whether there remains a `state avenue open to process such a claim' in this case." Id., at 768. Four dissenting judges argued the court should have reached the merits, rather than certifying questions to the state court; these judges asserted that House could not obtain relief under Schlup, let alone Sawyer and Herrera. 311 F. 3d, at 780-781 (opinion of Boggs, J.). A fifth dissenter explained that while he agreed with the majority that House "presents a strong claim for habeas relief, at least at the sentencing phase of the case," he objected to the certification of questions to the Tennessee high court. Id., at 787 (opinion of Gilman, J.). This Court denied certiorari. Bell v. House, 539 U.S. 937 (2003).

The State urged the Tennessee Supreme Court not to answer the Court of Appeals' certified questions, and the state court did not do so. The case returned to the United States Court of Appeals for the Sixth Circuit. This time an eight-judge majority affirmed the District Court's denial of habeas *536 relief. 386 F.3d 668 (2004). Six dissenters argued that House not only had met the actual-innocence standard for overcoming procedural default but also was entitled to immediate release under Herrera. 386 F. 3d, at 708 (opinion of Merritt, J.). A seventh dissenter (the same judge who wrote separately in the previous en banc decision) described the case as "a real-life murder mystery, an authentic `who-done-it' where the wrong man may be executed." Id., at 709 (opinion of Gilman, J.). He concluded such grave uncertainty necessitated relief in the form of a new trial for House. Id., at 710.

We granted certiorari, 545 U.S. 1151 (2005), and now reverse.

IV

As a general rule, claims forfeited under state law may support federal habeas relief only if the prisoner demonstrates cause for the default and prejudice from the asserted error. See Murray v. Carrier, 477 U.S. 478, 485 (1986); Engle v. Isaac, 456 U.S. 107, 129 (1982); Wainwright v. Sykes, 433 U.S. 72, 87 (1977). The rule is based on the comity and respect that must be accorded to state-court judgments. See, e. g., Engle, supra, at 126-129; Wainwright, supra, at 89-90. The bar is not, however, unqualified. In an effort to "balance the societal interests in finality, comity, and conservation of scarce judicial resources with the individual interest in justice that arises in the extraordinary case," Schlup, 513 U. S., at 324, the Court has recognized a miscarriage-of-justice exception. "`[I]n appropriate cases,'" the Court has said, "the principles of comity and finality that inform the concepts of cause and prejudice `must yield to the imperative of correcting a fundamentally unjust incarceration,'" Carrier, supra, at 495 (quoting Engle, supra, at 135).

In Schlup, the Court adopted a specific rule to implement this general principle. It held that prisoners asserting innocence as a gateway to defaulted claims must establish that, *537 in light of new evidence, "it is more likely than not that no reasonable juror would have found petitioner guilty beyond a reasonable doubt." 513 U.S., at 327. This formulation, Schlup explains, "ensures that petitioner's case is truly `extraordinary,' while still providing petitioner a meaningful avenue by which to avoid a manifest injustice." Ibid. (quoting McCleskey v. Zant, 499 U.S. 467, 494 (1991)). In the usual case the presumed guilt of a prisoner convicted in state court counsels against federal review of defaulted claims. Yet a petition supported by a convincing Schlup gateway showing "raise[s] sufficient doubt about [the petitioner's] guilt to undermine confidence in the result of the trial without the assurance that that trial was untainted by constitutional error"; hence, "a review of the merits of the constitutional claims" is justified. 513 U.S., at 317.

For purposes of this case several features of the Schlup standard bear emphasis. First, although "[t]o be credible" a gateway claim requires "new reliable evidence—whether it be exculpatory scientific evidence, trustworthy eyewitness accounts, or critical physical evidence—that was not presented at trial," id., at 324, the habeas court's analysis is not limited to such evidence. There is no dispute in this case that House has presented some new reliable evidence; the State has conceded as much, see infra, at 540-541. In addition, because the District Court held an evidentiary hearing in this case, and because the State does not challenge the court's decision to do so, we have no occasion to elaborate on Schlup's observation that when considering an actual-innocence claim in the context of a request for an evidentiary hearing, the District Court need not "test the new evidence by a standard appropriate for deciding a motion for summary judgment," but rather may "consider how the timing of the submission and the likely credibility of the affiants bear on the probable reliability of that evidence." 513 U.S., at 331-332. Our review in this case addresses the merits of the Schlup inquiry, based on a fully developed record, and with *538 respect to that inquiry Schlup makes plain that the habeas court must consider "`all the evidence,'" old and new, incriminating and exculpatory, without regard to whether it would necessarily be admitted under "rules of admissibility that would govern at trial." See id., at 327-328 (quoting Friendly, Is Innocence Irrelevant? Collateral Attack on Criminal Judgments, 38 U. Chi. L. Rev. 142, 160 (1970)). Based on this total record, the court must make "a probabilistic determination about what reasonable, properly instructed jurors would do." 513 U.S., at 329. The court's function is not to make an independent factual determination about what likely occurred, but rather to assess the likely impact of the evidence on reasonable jurors. Ibid.

Second, it bears repeating that the Schlup standard is demanding and permits review only in the "`extraordinary'" case. Id., at 327 (quoting Zant, supra, at 494); see also 513 U.S., at 324 (emphasizing that "in the vast majority of cases, claims of actual innocence are rarely successful"). At the same time, though, the Schlup standard does not require absolute certainty about the petitioner's guilt or innocence. A petitioner's burden at the gateway stage is to demonstrate that more likely than not, in light of the new evidence, no reasonable juror would find him guilty beyond a reasonable doubt—or, to remove the double negative, that more likely than not any reasonable juror would have reasonable doubt.

Finally, as the Schlup decision explains, the gateway actual-innocence standard is "by no means equivalent to the standard of Jackson v. Virginia, 443 U.S. 307 (1979)," which governs claims of insufficient evidence. Id., at 330. When confronted with a challenge based on trial evidence, courts presume the jury resolved evidentiary disputes reasonably so long as sufficient evidence supports the verdict. Because a Schlup claim involves evidence the trial jury did not have before it, the inquiry requires the federal court to assess how reasonable jurors would react to the overall, newly supplemented record. See ibid. If new evidence so requires, this *539 may include consideration of "the credibility of the witnesses presented at trial." Ibid.; see also ibid. (noting that "[i]n such a case, the habeas court may have to make some credibility assessments").

As an initial matter, the State argues that the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), 110 Stat. 1214, has replaced the Schlup standard with a stricter test based on Sawyer, which permits consideration of successive, abusive, or defaulted sentencing-related claims only if the petitioner "show[s] by clear and convincing evidence that, but for a constitutional error, no reasonable juror would have found the petitioner eligible for the death penalty under the applicable state law," 505 U.S., at 336. One AEDPA provision establishes a similar standard for second or successive petitions involving no retroactively applicable new law, 28 U.S. C. § 2244(b)(2)(B)(ii); another sets it as a threshold for obtaining an evidentiary hearing on claims the petitioner failed to develop in state court, § 2254(e)(2). Neither provision addresses the type of petition at issue here— a first federal habeas petition seeking consideration of defaulted claims based on a showing of actual innocence. Thus, the standard of review in these provisions is inapplicable. Cf. Lonchar v. Thomas, 517 U.S. 314, 324 (1996) ("Dismissal of a first federal habeas petition is a particularly serious matter").

The State also argues that the District Court's findings in this case tie our hands, precluding a ruling in House's favor absent a showing of clear error as to the District Court's specific determinations. This view overstates the effect of the District Court's ruling. Deference is given to a trial court's assessment of evidence presented to it in the first instance. Yet the Schlup inquiry, we repeat, requires a holistic judgment about "`all the evidence,'" 513 U.S., at 328 (quoting Friendly, supra, at 160), and its likely effect on reasonable jurors applying the reasonable-doubt standard. As a general rule, the inquiry does not turn on discrete findings *540 regarding disputed points of fact, and "[i]t is not the district court's independent judgment as to whether reasonable doubt exists that the standard addresses," 513 U.S., at 329. Here, although the District Court attentively managed complex proceedings, carefully reviewed the extensive record, and drew certain conclusions about the evidence, the court did not clearly apply Schlup's predictive standard regarding whether reasonable jurors would have reasonable doubt. As we shall explain, moreover, we are uncertain about the basis for some of the District Court's conclusions—a consideration that weakens our reliance on its determinations.

With this background in mind we turn to the evidence developed in House's federal habeas proceedings.

DNA Evidence

First, in direct contradiction of evidence presented at trial, DNA testing has established that the semen on Mrs. Muncey's nightgown and panties came from her husband, Mr. Muncey, not from House. The State, though conceding this point, insists this new evidence is immaterial. At the guilt phase at least, neither sexual contact nor motive were elements of the offense, so in the State's view the evidence, or lack of evidence, of sexual assault or sexual advance is of no consequence. We disagree. In fact we consider the new disclosure of central importance.

From beginning to end the case is about who committed the crime. When identity is in question, motive is key. The point, indeed, was not lost on the prosecution, for it introduced the evidence and relied on it in the final guilt-phase closing argument. Referring to "evidence at the scene," the prosecutor suggested that House committed, or attempted to commit, some "indignity" on Mrs. Muncey that neither she "nor any mother on that road would want to do with Mr. House." 9 Tr. 1302-1303. Particularly in a case like this where the proof was, as the State Supreme Court observed, circumstantial, State v. House, 743 S. W. 2d, at 143, 144, we *541 think a jury would have given this evidence great weight. Quite apart from providing proof of motive, it was the only forensic evidence at the scene that would link House to the murder.

Law and society, as they ought to do, demand accountability when a sexual offense has been committed, so not only did this evidence link House to the crime; it likely was a factor in persuading the jury not to let him go free. At sentencing, moreover, the jury came to the unanimous conclusion, beyond a reasonable doubt, that the murder was committed in the course of a rape or kidnaping. The alleged sexual motivation relates to both those determinations. This is particularly so given that, at the sentencing phase, the jury was advised that House had a previous conviction for sexual assault.

A jury informed that fluids on Mrs. Muncey's garments could have come from House might have found that House trekked the nearly two miles to the victim's home and lured her away in order to commit a sexual offense. By contrast a jury acting without the assumption that the semen could have come from House would have found it necessary to establish some different motive, or, if the same motive, an intent far more speculative. When the only direct evidence of sexual assault drops out of the case, so, too, does a central theme in the State's narrative linking House to the crime. In that light, furthermore, House's odd evening walk and his false statements to authorities, while still potentially incriminating, might appear less suspicious.

Bloodstains

The other relevant forensic evidence is the blood on House's pants, which appears in small, even minute, stains in scattered places. As the prosecutor told the jury, they were stains that, due to their small size, "you or I might not detect[,] [m]ight not see, but which the FBI lab was able to find on [House's] jeans." App. 11. The stains appear inside the *542 right pocket, outside that pocket, near the inside button, on the left thigh and outside leg, on the seat of the pants, and on the right bottom cuff, including inside the pants. Due to testing by the FBI, cuttings now appear on the pants in several places where stains evidently were found. (The cuttings were destroyed in the testing process, and defense experts were unable to replicate the tests.) At trial, the government argued "nothing that the defense has introduced in this case explains what blood is doing on his jeans, all over [House's] jeans, that is scientifically, completely different from his blood." Id., at 105. House, though not disputing at this point that the blood is Mrs. Muncey's, now presents an alternative explanation that, if credited, would undermine the probative value of the blood evidence.

During House's habeas proceedings, Dr. Cleland Blake, an Assistant Chief Medical Examiner for the State of Tennessee and a consultant in forensic pathology to the TBI for 22 years, testified that the blood on House's pants was chemically too degraded, and too similar to blood collected during the autopsy, to have come from Mrs. Muncey's body on the night of the crime. The blood samples collected during the autopsy were placed in test tubes without preservative. Under such conditions, according to Dr. Blake, "you will have enzyme degradation. You will have different blood group degradation, blood marker degradation." Record, Doc. 275, p. 80 (hereinafter R275:80). The problem of decay, moreover, would have been compounded by the body's long exposure to the elements, sitting outside for the better part of a summer day. In contrast, if blood is preserved on cloth, "it will stay there for years," ibid.; indeed, Dr. Blake said he deliberately places blood drops on gauze during autopsies to preserve it for later testing. The blood on House's pants, judging by Agent Bigbee's tests, showed "similar deterioration, breakdown of certain of the named numbered enzymes" as in the autopsy samples. Id., at 110. "[I]f the victim's blood had spilled on the jeans while the victim was *543 alive and this blood had dried," Dr. Blake stated, "the deterioration would not have occurred," ibid., and "you would expect [the blood on the jeans] to be different than what was in the tube," id., at 113. Dr. Blake thus concluded the blood on the jeans came from the autopsy samples, not from Mrs. Muncey's live (or recently killed) body.

Other evidence confirms that blood did in fact spill from the vials. It appears the vials passed from Dr. Carabia, who performed the autopsy, into the hands of two local law enforcement officers, who transported it to the FBI, where Agent Bigbee performed the enzyme tests. The blood was contained in four vials, evidently with neither preservative nor a proper seal. The vials, in turn, were stored in a styrofoam box, but nothing indicates the box was kept cool. Rather, in what an evidence protocol expert at the habeas hearing described as a violation of proper procedure, the styrofoam box was packed in the same cardboard box as other evidence including House's pants (apparently in a paper bag) and other clothing (in separate bags). The cardboard box was then carried in the officers' car while they made the 10-hour journey from Tennessee to the FBI lab. Dr. Blake stated that blood vials in hot conditions (such as a car trunk in the summer) could blow open; and in fact, by the time the blood reached the FBI it had hemolyzed, or spoiled, due to heat exposure. By the time the blood passed from the FBI to a defense expert, roughly a vial and a half were empty, though Agent Bigbee testified he used at most a quarter of one vial. Blood, moreover, had seeped onto one corner of the styrofoam box and onto packing gauze inside the box below the vials.

In addition, although the pants apparently were packaged initially in a paper bag and FBI records suggest they arrived at the FBI in one, the record does not contain the paper bag but does contain a plastic bag with a label listing the pants and Agent Scott's name—and the plastic bag has blood on it. The blood appears in a forked streak roughly five inches long *544 and two inches wide running down the bag's outside front. Though testing by House's expert confirmed the stain was blood, the expert could not determine the blood's source. Speculations about when and how the blood got there add to the confusion regarding the origins of the stains on House's pants.

Faced with these indications of, at best, poor evidence control, the State attempted to establish at the habeas hearing that all blood spillage occurred after Agent Bigbee examined the pants. Were that the case, of course, then blood would have been detected on the pants before any spill—which would tend to undermine Dr. Blake's analysis and support using the bloodstains to infer House's guilt. In support of this theory the State put on testimony by a blood spatter expert who believed the "majority" of the stains were "transfer stains," that is, stains resulting from "wip[ing] across the surface of the pants" rather than seeping or spillage. App. 293-294. Regarding the spillage in the styrofoam box, the expert noted that yellow "Tennessee Crime Laboratory" tape running around the box and down all four sides did not line up when the bloodstains on the box's corner were aligned. The inference was that the FBI received the box from Tennessee authorities, opened it, and resealed it before the spillage occurred. Reinforcing this theory, Agent Bigbee testified that he observed no blood spillage in the styrofoam box and that had he detected such signs of evidence contamination, FBI policy would have required immediate return of the evidence.

In response House argued that even assuming the tape alignment showed spillage occurring after FBI testing, spillage on one or more earlier occasions was likely. In fact even the State's spatter expert declined to suggest the blood in the box and on the packing gauze accounted for the full vial and a quarter missing. And when the defense expert opened the box and discovered the spills, the bulk of the blood-caked gauze was located around and underneath the *545 half-full vial, which was also located near the stained corner. No gauze immediately surrounding the completely empty vial was stained. The tape, moreover, circled the box in two layers, one underneath the other, and in one spot the underlying layer stops cleanly at the lid's edge, as if cut with a razor, and does not continue onto the body of the box below. In House's view this clean cut suggests the double layers could not have resulted simply from wrapping the tape around twice, as the spatter expert claimed; rather, someone possessing Tennessee Crime Lab tape—perhaps the officers transporting the blood and pants—must have cut the box open and resealed it, possibly creating an opportunity for spillage. Supporting the same inference, a label on the box's lid lists both blood and vaginal secretions as the box's contents, though Agent Bigbee's records show the vaginal fluids arrived at the FBI in a separate envelope. Finally, cross-examination revealed that Agent Bigbee's practice did not always match the letter of FBI policy. Although Mrs. Muncey's bra and housecoat were packed together in a single bag, creating, according to Agent Bigbee, a risk of "cross contamination," id., at 286, he did not return them; nor did he note the discrepancy between the "[b]lood and vaginal secretions" label and the styrofoam box's actual contents, though he insisted his customary practice was to match labels with contents immediately upon opening an evidence box, id., at 287.

The State challenged Dr. Blake's scientific conclusions, and to do so it called Agent Bigbee as a witness. Agent Bigbee defended the testimony he had given at the trial. To begin with, he suggested Dr. Blake had misconstrued the term "inc" in Agent Bigbee's trial report, interpreting it to mean "incomplete" when it in fact meant "inconclusive." Id., at 254-256, 282. Dr. Blake, however, replied "[s]ame difference" when asked whether his opinion would change if "inc" meant "inconclusive." Id., at 256; see also 6 Tr. 906 (Bigbee trial testimony) ("You will notice I have INC written under *546 the transparent, that is the symbol that I use to mean the test was incomplete"). Agent Bigbee further asserted that, whereas Dr. Blake (in Bigbee's view) construed the results to mean the enzyme was not present at all, in fact the results indicated only that Bigbee could not identify the marker type on whatever enzymes were present. App. 282. Yet the State did not cross-examine Dr. Blake on this point, nor did the District Court resolve the dispute one way or the other, so on this record it seems possible that Dr. Blake meant only to suggest the blood was too degraded to permit conclusive typing. The State, moreover, does not ask us to question Dr. Blake's basic premise about the durability of blood chemicals deposited on cotton—a premise Agent Bigbee appeared to accept as a general matter. Given the record as it stands, then, we cannot say Dr. Blake's conclusions have been discredited; if other objections might be adduced, they must await further proceedings. At the least, the record before us contains credible testimony suggesting that the missing enzyme markers are generally better preserved on cloth than in poorly kept test tubes, and that principle could support House's spillage theory for the blood's origin.

In this Court, as a further attack on House's showing, the State suggests that, given the spatter expert's testimony, House's theory would require a jury to surmise that Tennessee officials donned the pants and deliberately spread blood over them. We disagree. This should be a matter for the trier of fact to consider in the first instance, but we can note a line of argument that could refute the State's position. It is correct that the State's spatter expert opined that the stains resulted from wiping or smearing rather than direct spillage; and she further stated that the distribution of stains in some spots suggests the pants were "folded in some manner or creased in some manner" when the transfers occurred, id., at 296. While the expert described this pattern, at least with respect to stains on the lap of the pants, as "consistent" with the pants being worn at the time of the staining, ibid., *547 her testimony, as we understand it, does not refute the hypothesis that the packaging of the pants for transport was what caused them to be folded or creased. It seems permissible, moreover, to conclude that the small size and wide distribution of stains—inside the right pocket, outside that pocket, near the inside button, on the left thigh and outside leg, on the seat of the pants, and on the right bottom cuff, including inside the pants—fits as well with spillage in transport as with wiping and smearing from bloody objects at the crime scene, as the State proposes. (As has been noted, no blood was found on House's shoes.)

The District Court discounted Dr. Blake's opinion, not on account of Blake's substantive approach, but based on testimony from Agent Scott indicating he saw, as the District Court put it, "what appeared to be bloodstains on Mr. House's blue jeans when the jeans were removed from the laundry hamper at Ms. Turner's trailer." Id., at 348. This inference seems at least open to question, however. Agent Scott stated only that he "saw reddish brownish stains [he] suspected to be blood"; he admitted that he "didn't thoroughly examine the blue jeans at that time." R276:113-114. The pants were in fact extensively soiled with mud and reddish stains, only small portions of which are blood.

In sum, considering "`all the evidence,'" Schlup, 513 U. S., at 328 (quoting Friendly, 38 U. Chi. L. Rev., at 160), on this issue, we think the evidentiary disarray surrounding the blood, taken together with Dr. Blake's testimony and the limited rebuttal of it in the present record, would prevent reasonable jurors from placing significant reliance on the blood evidence. We now know, though the trial jury did not, that an Assistant Chief Medical Examiner believes the blood on House's jeans must have come from autopsy samples; that a vial and a quarter of autopsy blood is unaccounted for; that the blood was transported to the FBI together with the pants in conditions that could have caused vials to spill; that the blood did indeed spill at least once during its journey *548 from Tennessee authorities through FBI hands to a defense expert; that the pants were stored in a plastic bag bearing both a large blood stain and a label with TBI Agent Scott's name; and that the styrofoam box containing the blood samples may well have been opened before it arrived at the FBI lab. Thus, whereas the bloodstains, emphasized by the prosecution, seemed strong evidence of House's guilt at trial, the record now raises substantial questions about the blood's origin.

A Different Suspect

Were House's challenge to the State's case limited to the questions he has raised about the blood and semen, the other evidence favoring the prosecution might well suffice to bar relief. There is, however, more; for in the post-trial proceedings House presented troubling evidence that Mr. Muncey, the victim's husband, himself could have been the murderer.

At trial, as has been noted, the jury heard that roughly two weeks before the murder Mrs. Muncey's brother received a frightened phone call from his sister indicating that she and Mr. Muncey had been fighting, that she was scared, and that she wanted to leave him. The jury also learned that the brother once saw Mr. Muncey "smac[k]" the victim. 7 Tr. 1087-1088. House now has produced evidence from multiple sources suggesting that Mr. Muncey regularly abused his wife. For example, one witness—Kathy Parker, a lifelong area resident who denied any animosity toward Mr. Muncey—recalled that Mrs. Muncey "was constantly with black eyes and busted mouth." App. 235. In addition Hazel Miller, who is Kathy Parker's mother and a lifelong acquaintance of Mr. Muncey, testified at the habeas hearing that two or three months before the victim's death Mr. Muncey came to Miller's home and "tried to get my daughter [Parker] to go out with him," R274:47. (Parker had dated Mr. Muncey at age 14.) According to Miller, Muncey said "[h]e was upset with his wife, that they had had an argument *549 and he said he was going to get rid of that woman one way or the other." App. 236.

Another witness—Mary Atkins, also an area native who "grew up" with Mr. Muncey and professed no hard feelings, R274:10, 16—claims she saw Mr. Muncey "backhan[d]" Mrs. Muncey on the very night of the murder. App. 226, 228. Atkins recalled that during a break in the recreation center dance, she saw Mr. Muncey and his wife arguing in the parking lot. Mr. Muncey "grabbed her and he just backhanded her." Id., at 228. After that, Mrs. Muncey "left walking." Id., at 229. There was also testimony from Atkins' mother, named Artie Lawson. A self-described "good friend" of Mr. Muncey, id., at 231, Lawson said Mr. Muncey visited her the morning after the murder, before the body was found. According to Lawson, Mr. Muncey asked her to tell anyone who inquired not only that she had been at the dance the evening before and had seen him, but also that he had breakfasted at her home at 6 o'clock that morning. Lawson had not in fact been at the dance, nor had Mr. Muncey been with her so early.

Of most importance is the testimony of Kathy Parker and her sister Penny Letner. They testified at the habeas hearing that, around the time of House's trial, Mr. Muncey had confessed to the crime. Parker recalled that she and "some family members and some friends [were] sitting around drinking" at Parker's trailer when Mr. Muncey "just walked in and sit down." R274:37. Muncey, who had evidently been drinking heavily, began "rambling off ... [t]alking about what happened to his wife and how it happened and he didn't mean to do it." Ibid. According to Parker, Mr. Muncey "said they had been into [an] argument and he slapped her and she fell and hit her head and it killed her and he didn't mean for it to happen." Id., at 38. Parker said she "freaked out and run him off." Ibid.

Letner similarly recalled that at some point either "during [House's] trial or just before," id., at 30, Mr. Muncey intruded *550 on a gathering at Parker's home. Appearing "pretty well blistered," Muncey "went to crying and was talking about his wife and her death and he was saying that he didn't mean to do it." App. 232. "[D]idn't mean to do what[?]," Letner asked, R274:33, at which point Mr. Muncey explained:

"[S]he was `bitching him out' because he didn't take her fishing that night, that he went to the dance instead. He said when he come home that she was still on him pretty heavily `bitching him out' again and that he smacked her and that she fell and hit her head. He said I didn't mean to do it, but I had to get rid of her, because I didn't want to be charged with murder." App. 232-233.

Letner, who was then 19 years old with a small child, said Mr. Muncey's statement "scared [her] quite badly," so she "got out of there immediately." Id., at 233. Asked whether she reported the incident to the authorities, Letner stated, "I was frightened, you know.... I figured me being 19 year old they wouldn't listen to anything I had to say." R274:31. Parker, on the other hand, claimed she (Parker) in fact went to the Sheriff's Department, but no one would listen:

"I tried to speak to the Sheriff but he was real busy. He sent me to a deputy. The deputy told me to go upstairs to the courtroom and talk to this guy, I can't remember his name. I never did really get to talk to anybody." App. 234.

Parker said she did not discuss the matter further because "[t]hey had it all signed, sealed and delivered. We didn't know anything to do until we heard that they reopened [House's] trial." R274:45. Parker's mother, Hazel Miller, confirmed she had driven Parker to the courthouse, where Parker "went to talk to some of the people about this case." App. 237.

Other testimony suggests Mr. Muncey had the opportunity to commit the crime. According to Dennis Wallace, a local *551 law enforcement official who provided security at the dance on the night of the murder, Mr. Muncey left the dance "around 10:00, 10:30, 9:30 to 10:30." R274:56-57. Although Mr. Muncey told law enforcement officials just after the murder that he left the dance only briefly and returned, Wallace could not recall seeing him back there again. Later that evening, Wallace responded to Mr. Muncey's report that his wife was missing. Muncey denied he and his wife had been "a fussing or a fighting"; he claimed his wife had been "kidnapped." Id., at 58. Wallace did not recall seeing any blood, disarray, or knocked-over furniture, although he admitted he "didn't pay too much attention" to whether the floor appeared especially clean. According to Wallace, Mr. Muncey said "let's search for her" and then led Wallace out to search "in the weeds" around the home and the driveway (not out on the road where the body was found). Id., at 58, 60, 63.

In the habeas proceedings, then, two different witnesses (Parker and Letner) described a confession by Mr. Muncey; two more (Atkins and Lawson) described suspicious behavior (a fight and an attempt to construct a false alibi) around the time of the crime; and still other witnesses described a history of abuse.

As to Parker and Letner, the District Court noted that it was "not impressed with the allegations of individuals who wait over ten years to come forward with their evidence," especially considering that "there was no physical evidence in the Munceys' kitchen to corroborate [Mr. Muncey's] alleged confession that he killed [his wife] there." App. 348. Parker and Letner, however, did attempt to explain their delay coming forward, and the record indicates no reason why these two women, both lifelong acquaintances of Mr. Muncey, would have wanted either to frame him or to help House. Furthermore, the record includes at least some independent support for the statements Parker and Letner attributed to Mr. Muncey. The supposed explanation for the *552 fatal fight—that his wife was complaining about going fishing—fits with Mrs. Muncey's statement to Luttrell earlier that evening that her husband's absence was "all right, because she was going to make him take her fishing the next day," id., at 11-12. And Dr. Blake testified, in only partial contradiction of Dr. Carabia, that Mrs. Muncey's head injury resulted from "a surface with an edge" or "a hard surface with a corner," not from a fist. R275:72. (Dr. Carabia had said either a fist or some other object could have been the cause.)

Mr. Muncey testified at the habeas hearing, and the District Court did not question his credibility. Though Mr. Muncey said he seemed to remember visiting Lawson the day after the murder, he denied either killing his wife or confessing to doing so. Yet Mr. Muncey also claimed, contrary to Constable Wallace's testimony and to his own prior statement, that he left the dance on the night of the crime only when it ended at midnight. Mr. Muncey, moreover, denied ever hitting Mrs. Muncey; the State itself had to impeach him with a prior statement on this point.

It bears emphasis, finally, that Parker's and Letner's testimony is not comparable to the sort of eleventh-hour affidavit vouching for a defendant and incriminating a conveniently absent suspect that Justice O'Connor described in her concurring opinion in Herrera as "unfortunate" and "not uncommon" in capital cases, 506 U.S., at 423; nor was the confession Parker and Letner described induced under pressure of interrogation. The confession evidence here involves an alleged spontaneous statement recounted by two eyewitnesses with no evident motive to lie. For this reason it has more probative value than, for example, incriminating testimony from inmates, suspects, or friends or relations of the accused.

The evidence pointing to Mr. Muncey is by no means conclusive. If considered in isolation, a reasonable jury might well disregard it. In combination, however, with the challenges *553 to the blood evidence and the lack of motive with respect to House, the evidence pointing to Mr. Muncey likely would reinforce other doubts as to House's guilt.

Other Evidence

Certain other details were presented at the habeas hearing. First, Dr. Blake, in addition to testifying about the blood evidence and the victim's head injury, examined photographs of House's bruises and scratches and concluded, based on 35 years' experience monitoring the development and healing of bruises, that they were too old to have resulted from the crime. In addition Dr. Blake claimed that the injury on House's right knuckle was indicative of "[g]etting mashed"; it was not consistent with striking someone. R275:63. (That of course would also eliminate the explanation that the injury came from the blow House supposedly told Turner he gave to his unidentified assailant.)

The victim's daughter, Lora Muncey (now Lora Tharp), also testified at the habeas hearing. She repeated her recollection of hearing a man with a deep voice like her grandfather's and a statement that her father had had a wreck down by the creek. She also denied seeing any signs of struggle or hearing a fight between her parents, though she also said she could not recall her parents ever fighting physically. The District Court found her credible, and this testimony certainly cuts in favor of the State.

Finally, House himself testified at the habeas proceedings. He essentially repeated the story he allegedly told Turner about getting attacked on the road. The District Court found, however, based on House's demeanor, that he "was not a credible witness." App. 329.

Conclusion

This is not a case of conclusive exoneration. Some aspects of the State's evidence—Lora Muncey's memory of a deep voice, House's bizarre evening walk, his lie to law enforcement, *554 his appearance near the body, and the blood on his pants—still support an inference of guilt. Yet the central forensic proof connecting House to the crime—the blood and the semen—has been called into question, and House has put forward substantial evidence pointing to a different suspect. Accordingly, and although the issue is close, we conclude that this is the rare case where—had the jury heard all the conflicting testimony—it is more likely than not that no reasonable juror viewing the record as a whole would lack reasonable doubt.

V

In addition to his gateway claim under Schlup, House argues that he has shown freestanding innocence and that as a result his imprisonment and planned execution are unconstitutional. In Herrera, decided three years before Schlup, the Court assumed without deciding that "in a capital case a truly persuasive demonstration of `actual innocence' made after trial would render the execution of a defendant unconstitutional, and warrant federal habeas relief if there were no state avenue open to process such a claim." 506 U.S., at 417; see also id., at 419 (O'Connor, J., concurring) ("I cannot disagree with the fundamental legal principle that executing the innocent is inconsistent with the Constitution"). "[T]he threshold showing for such an assumed right would necessarily be extraordinarily high," the Court explained, and petitioner's evidence there fell "far short of that which would have to be made in order to trigger the sort of constitutional claim which we have assumed, arguendo, to exist." Id., at 417, 418-419; see also id., at 427 (O'Connor, J., concurring) (noting that because "[p]etitioner has failed to make a persuasive showing of actual innocence," "the Court has no reason to pass on, and appropriately reserves, the question whether federal courts may entertain convincing claims of actual innocence"). House urges the Court to answer the question left open in Herrera and hold not only that free-standing *555 innocence claims are possible but also that he has established one.

We decline to resolve this issue. We conclude here, much as in Herrera, that whatever burden a hypothetical freestanding innocence claim would require, this petitioner has not satisfied it. To be sure, House has cast considerable doubt on his guilt—doubt sufficient to satisfy Schlup's gateway standard for obtaining federal review despite a state procedural default. In Herrera, however, the Court described the threshold for any hypothetical freestanding innocence claim as "extraordinarily high." 506 U.S., at 417. The sequence of the Court's decisions in Herrera and Schlup—first leaving unresolved the status of freestanding claims and then establishing the gateway standard—implies at the least that Herrera requires more convincing proof of innocence than Schlup. It follows, given the closeness of the Schlup question here, that House's showing falls short of the threshold implied in Herrera.

* * *

House has satisfied the gateway standard set forth in Schlup and may proceed on remand with procedurally defaulted constitutional claims. The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

JUSTICE ALITO took no part in the consideration or decision of this case.

CHIEF JUSTICE ROBERTS, with whom Justice SCALIA and JUSTICE THOMAS join, concurring in the judgment in part and dissenting in part.

To overcome the procedural hurdle that Paul House created by failing to properly present his constitutional claims to a Tennessee court, he must demonstrate that the constitutional violations he alleges "ha[ve] probably resulted in the *556 conviction of one who is actually innocent," such that a federal court's refusal to hear the defaulted claims would be a "miscarriage of justice." Schlup v. Delo, 513 U.S. 298, 326, 327 (1995) (internal quotation marks omitted). To make the requisite showing of actual innocence, House must produce "new reliable evidence" and "must show that it is more likely than not that no reasonable juror would have convicted him in the light of the new evidence." Id., at 324, 327 (emphasis added). The question is not whether House was prejudiced at his trial because the jurors were not aware of the new evidence, but whether all the evidence, considered together, proves that House was actually innocent, so that no reasonable juror would vote to convict him. Considering all the evidence, and giving due regard to the District Court's findings on whether House's new evidence was reliable, I do not find it probable that no reasonable juror would vote to convict him, and accordingly I dissent.

Because I do not think that House has satisfied the actual innocence standard set forth in Schlup, I do not believe that he has met the higher threshold for a freestanding innocence claim, assuming such a claim exists. See Herrera v. Collins, 506 U.S. 390, 417 (1993). I therefore concur in the judgment with respect to the Court's disposition of that separate claim.

I

In Schlup, we stated that a habeas petitioner attempting to present a defaulted claim to a federal court must present "new reliable evidence—whether it be exculpatory scientific evidence, trustworthy eyewitness accounts, or critical physical evidence—that was not presented at trial." 513 U.S., at 324 (emphasis added). Implicit in the requirement that a habeas petitioner present reliable evidence is the expectation that a factfinder will assess reliability. The new evidence at issue in Schlup had not been subjected to such an assessment—the claim in Schlup was for an evidentiary hearing— *557 and this Court specifically recognized that the "new statements may, of course, be unreliable." Id., at 331. The Court stated that the District Court, as the "reviewing tribunal," was tasked with assessing the "probative force" of the petitioner's new evidence of innocence, and "may have to make some credibility assessments." Id., at 327-328, 330. Indeed, the Supreme Court took the unusual step of remanding the case to the Court of Appeals "with instructions to remand to the District Court," so that the District Court could consider how the "likely credibility of the affiants" bears upon the "probable reliability" of the new evidence. Id., at 332. In short, the new evidence is not simply taken at face value; its reliability has to be tested.

Critical to the Court's conclusion here that House has sufficiently demonstrated his innocence are three pieces of new evidence presented to the District Court: DNA evidence showing that the semen on Carolyn Muncey's clothing was from her husband, Hubert Muncey, not from House; testimony from new witnesses implicating Mr. Muncey in the murder; and evidence indicating that Mrs. Muncey's blood spilled from test tubes containing autopsy samples in an evidence container. To determine whether it should open its door to House's defaulted constitutional claims, the District Court considered this evidence in a comprehensive evidentiary hearing. As House presented his new evidence, and as the State rebutted it, the District Court observed the witnesses' demeanor, examined physical evidence, and made findings about whether House's new evidence was in fact reliable. This factfinding role is familiar to a district court. "The trial judge's major role is the determination of fact, and with experience in fulfilling that role comes expertise." Anderson v. Bessemer City, 470 U.S. 564, 574 (1985).

The State did not contest House's new DNA evidence excluding him as the source of the semen on Mrs. Muncey's clothing, but it strongly contested the new testimony implicating *558 Mr. Muncey, and it insisted that the blood spillage occurred after the FBI tested House's jeans and determined that they were stained with Mrs. Muncey's blood.

At the evidentiary hearing, sisters Kathy Parker and Penny Letner testified that 14 years earlier, either during or around the time of House's trial, they heard Mr. Muncey drunkenly confess to having accidentally killed his wife when he struck her in their home during an argument, causing her to fall and hit her head. Record, Doc. 274, pp. 28-29, 30, 37-38. Schlup provided guidance on how a district court should assess this type of new evidence: The court "may consider how the timing of the submission and the likely credibility of the affiants bear on the probable reliability of that evidence," and it "must assess the probative force of the newly presented evidence in connection with the evidence of guilt adduced at trial." 513 U.S., at 332. Consistent with this guidance, the District Court concluded that the sisters' testimony was not credible. The court noted that it was "not impressed with the allegations of individuals who wait over ten years to come forward." App. 348. It also considered how the new testimony fit within the larger web of evidence, observing that Mr. Muncey's alleged confession contradicted the testimony of the Munceys' "very credible" daughter, Lora Tharp, who consistently testified that she did not hear a fight in the house that night, but instead heard a man with a deep voice who lured her mother from the house by saying that Mr. Muncey had been in a wreck near the creek. Id., at 323, 348.

The District Court engaged in a similar reliability inquiry with regard to House's new evidence of blood spillage. At the evidentiary hearing, House conceded that FBI testing showed that his jeans were stained with Mrs. Muncey's blood, but he set out to prove that the blood spilled from test tubes containing autopsy samples, and that it did so before the jeans were tested by the FBI. The District Court summarized the testimony of the various witnesses who handled *559 the evidence and their recollections about bloodstains and spillage; it acknowledged that House's expert, Dr. Cleland Blake, disagreed with FBI Agent Paul Bigbee about how to interpret the results of Agent Bigbee's genetic marker analysis summary; and it summarized the testimony of the State's blood spatter expert, Paulette Sutton. Id., at 339-347. After reviewing all the evidence, the District Court stated: "Based upon the evidence introduced during the evidentiary hearing . . . the court concludes that the spillage occurred after the FBI crime laboratory received and tested the evidence." Id., at 348 (emphasis added).

Normally, an appellate court reviews a district court's factual findings only for clear error. See Fed. Rule Civ. Proc. 52(a) ("Findings of fact, whether based on oral or documentary evidence, shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses"); Bessemer City, supra, at 574 (clearly-erroneous standard applies "even when the district court's findings do not rest on credibility determinations, but are based instead on physical or documentary evidence or inferences from other facts"). The Sixth Circuit deferred to the District Court's factual findings, 386 F.3d 668, 684 (2004), and Schlup did not purport to alter—but instead reaffirmed and highlighted—the district court's critical role as factfinder. Yet the majority asserts that the clear error standard "overstates the effect of the District Court's ruling," and then dismisses the District Court's reliability findings because it is "uncertain about" them, while stopping short of identifying clear error. Ante, at 539-540. This is a sharp departure from the guidance in Schlup.

In Schlup, we contrasted a district court's role in assessing the reliability of new evidence of innocence with a district court's role in deciding a summary judgment motion. 513 U.S., at 332. We explained that, in the latter situation, the district court does not assess credibility or weigh the evidence, *560 but simply determines whether there is a genuine factual issue for trial. Ibid. Assessing the reliability of new evidence, on the other hand, is a typical factfinding role, requiring credibility determinations and a weighing of the "probative force" of the new evidence in light of "the evidence of guilt adduced at trial." Ibid. We found it "[o]bviou[s]" that a habeas court conducting an actual innocence inquiry must do more than simply check whether there are genuine factual issues for trial. Ibid. The point of the actual innocence inquiry is for the federal habeas court to satisfy itself that it should suspend the normal procedural default rule, disregard the important judicial interests of finality and comity, and allow a state prisoner to present his defaulted constitutional claims to a federal court. See McCleskey v. Zant, 499 U.S. 467, 490-491 (1991).

The majority surprisingly states that this guidance is inapplicable here because this case involves a "fully developed record," while the District Court in Schlup had declined to conduct an evidentiary hearing. Ante, at 537-538. But the guidance is clearly applicable: The point in Schlup was not simply that a hearing was required, but why—because the District Court had to assess the probative force of the petitioner's newly presented evidence, by engaging in factfinding rather than performing a summary-judgment-type inquiry. 513 U.S., at 331-332. That is precisely what the District Court did here. In addition to a "fully developed record," we have the District Court's factual findings about the reliability of the new evidence in that record, factual findings which the majority disregards without finding clear error.

The majority essentially disregards the District Court's role in assessing the reliability of House's new evidence. With regard to the sisters' testimony, the majority casts aside the District Court's determination that their statements came too late and were too inconsistent with credible record evidence to be reliable, instead observing that the women had no obvious reason to lie, that a few aspects of *561 their testimony have record support, and that they recounted an uncoerced confession. Ante, at 551-552. As for the District Court's express finding that the autopsy blood spilled after the FBI tested House's jeans, the majority points to Dr. Blake's testimony that blood enzymes "are generally better preserved on cloth," and even conjures up its own theory in an attempt to refute Ms. Sutton's expert testimony that the pattern of some bloodstains was consistent with blood being transferred while the pants were being worn. Ante, at 546-547 ("This should be a matter for the trier of fact to consider in the first instance, but we can note a line of argument that could refute the State's position. . . . [Ms. Sutton's] testimony . . . does not refute the hypothesis that the packaging of the pants for transport was what caused them to be folded or creased"); see App. 296.

The majority's assessment of House's new evidence is precisely the summary-judgment-type inquiry Schlup said was inappropriate. 513 U.S., at 332. By casting aside the District Court's factual determinations made after a comprehensive evidentiary hearing, the majority has done little more than reiterate the factual disputes presented below. Witnesses do not testify in our courtroom, and it is not our role to make credibility findings and construct theories of the possible ways in which Mrs. Muncey's blood could have been spattered and wiped on House's jeans. The District Court did not painstakingly conduct an evidentiary hearing to compile a record for us to sort through transcript by transcript and photograph by photograph, assessing for ourselves the reliability of what we see. Schlup made abundantly clear that reliability determinations were essential, but were for the district court to make. Id., at 331-332. We are to defer to the better situated District Court on reliability, unless we determine that its findings are clearly erroneous. We are not concerned with "the district court's independent judgment as to whether reasonable doubt exists," id., at 329, but the District Court here made basic factual findings about *562 the reliability of House's new evidence; it did not offer its personal opinion about whether it doubted House's guilt. Schlup makes clear that those findings are controlling unless clearly erroneous.

I have found no clear error in the District Court's reliability findings. Not having observed Ms. Parker and Ms. Letner testify, I would defer to the District Court's determination that they are not credible, and the evidence in the record undermining the tale of an accidental killing during a fight in the Muncey home convinces me that this credibility finding is not clearly erroneous. Dr. Alex Carabia, who performed the autopsy, testified to injuries far more severe than a bump on the head: Mrs. Muncey had bruises on the front and back of her neck, on both thighs, on her lower right leg and left knee, and her hands were bloodstained up to the wrists; her injuries were consistent with a struggle and traumatic strangulation. Record, Addendum 4, 7 Tr. of Evidence in No. 378 (Crim. Ct. Union County, Tenn.), pp. 984-987 (hereinafter Tr.). And, of course, Lora Tharp has consistently recalled a deep-voiced visitor arriving late at night to tell Mrs. Muncey that her husband was in a wreck near the creek. App. 19, 270.

I also find abundant evidence in the record to support the District Court's finding that blood spilled within the evidence container after the FBI received and tested House's jeans. Agent Bigbee testified that there was no leakage in the items submitted to him for testing. Id., at 277. The majority's entire analysis on this point assumes the agent flatly lied, though there was no attack on his credibility below. Moreover, Ms. Sutton determined, in her expert opinion, that the wide distribution of stains "front and back, top to bottom," the fact that some bloodstains were mixed with mud, and the presence of bloodstains inside the pocket and inside the fly, showed that the blood was spattered and wiped—not spilled—on House's jeans. Id., at 291-293, 295; id., at 293 ("[I]f a tube of blood had spilled on these pants, *563 the stain should have been in a localized area"); id., at 294 ("The stains also . . . either originate on the inside and don't soak out or on the outside and are not soaking to the inside. That, of course, would be what you would see with a spill").

It is also worth noting that the blood evidently spilled inside the evidence container when the jeans were protected inside a plastic zip lock bag, as shown by the presence of a bloodstain on the outside of that bag. See Record, Pl. Exh. 10-6. House's expert tested the exterior and interior of that plastic bag for bloodstains using an "extremely sensitive" test, and only the exterior of the bag tested positive for blood. Id., Doc. 274, at 95-96. The evidence in the record indicates that the jeans were placed in the plastic bag after they arrived at the FBI: FBI records show that the jeans arrived there in a paper bag, and the plastic bag has FBI markings on it. Id., Addendum 2, Trial Exh. 31, at 36; id., Pl. Exh. 10-6. The bloodstain on the outside of the plastic bag therefore further supports the District Court's conclusion that the blood spilled after the evidence was received and tested by the FBI, and not en route when the jeans were in a paper bag. I suppose it is theoretically possible that the jeans were contaminated by spillage before arriving at the FBI, that Agent Bigbee either failed to note or lied about such spillage, and that the FBI then transferred the jeans into a plastic bag and put them back inside the evidence container with the spilled blood still sloshing around sufficiently to contaminate the outside of the plastic bag as extensively as it did. This sort of unbridled speculation can theoretically defeat any inconvenient fact, but does not suffice to convince me that the District Court's factual finding—that the blood spilled after FBI testing—was clearly erroneous.

Moreover, the yellow "Tennessee Crime Lab" tape placed around the container on all four sides does not line up when the bloodstained corners of the container and its lid are aligned, showing that the blood did not spill until sometime after the container was received and opened at its first destination— *564 the FBI. See id., Respondent's Exh. 24; id., Doc. 276, at 190-191 (testimony of Paulette Sutton). The majority points out that on one side of the container, the first of two layers of tape appears to begin cleanly at the lid's edge, and from this concludes that the container must have been cut open and resealed by Tennessee authorities en route to the FBI. Ante, at 545; see Record, Respondent's Exh. 23d. Even if the majority's deduction from a photograph of the container were true, it would show only that Tennessee authorities had reason to open the container once it was sealed to take something out or put something in, perhaps back at the crime lab in Union County. But even if the container had been opened before its arrival at the FBI, the majority recognizes that it was resealed with "Tennessee Crime Lab" tape, and the second layer of tape aligns only when the bloodstains on the container and its lid do not. Ante, at 544. Of course, the District Court—which concluded that the blood was spilled after testing at the FBI laboratory—had before it the box itself with the tape as the witnesses testified on the point, and not—like this Court—simply a photograph. See Bessemer City, 470 U. S., at 574 (district court's findings about physical evidence are reviewed for clear error).

House's theory that the blood on his jeans was transferred there from the autopsy samples is based on Dr. Blake's reading of Agent Bigbee's enzyme marker analysis summary. After reading the summary, Dr. Blake concluded that the enzymes in the bloodstains on House's jeans and the enzymes in the autopsy samples had deteriorated to the same extent. Record, Doc. 275, at 110. In particular, he noted that the GLO1 enzyme showed "incomplete penetration" on both the autopsy blood and the jeans, and because enzymes are better preserved on cloth, the enzyme should have been present on the jeans. Id., at 116. But Agent Bigbee disputed Dr. Blake's reading of what was, after all, Agent Bigbee's own study. He testified that "`inc'" on his chart meant "inconclusive," *565 not "incomplete penetration," and that the term "inconclusive" meant that the enzyme was present, but could not be grouped into an ABO bloodtype. Id., Doc. 276, at 140. While pointing out that his summary showed different levels of enzymes in the two samples, Agent Bigbee also noted that many different factors—such as heat, dirt, or bacteria in a clothes hamper—could cause enzymes to degrade on cloth. Id., at 139, 167-170. Considering how House's new blood spillage evidence fits within the record as a whole, I can see no clear error in the District Court's express finding that the blood spilled in the evidence container after the FBI found Mrs. Muncey's blood on House's jeans.

The District Court attentively presided over a complex evidentiary hearing, often questioning witnesses extensively during the presentation of critical evidence. See, e. g., id., Doc. 275, at 110-115. The court concisely summarized the evidence presented, then dutifully made findings about the reliability of the testimony it heard and the evidence it observed. We are poorly equipped to second-guess the District Court's reliability findings and should defer to them, consistent with the guidance we provided in Schlup.

II

With due regard to the District Court's reliability findings, this case invites a straightforward application of the legal standard adopted in Schlup. A petitioner does not pass through the Schlup gateway if it is "more likely than not that there is any juror who, acting reasonably, would have found the petitioner guilty beyond a reasonable doubt." 513 U.S., at 333 (O'Connor, J., concurring) (emphasis added).

The majority states that if House had presented just one of his three key pieces of evidence—or even two of the three—he would not pass through the Schlup gateway. See ante, at 548 ("Were House's challenge to the State's case limited to the questions he has raised about the blood and semen, the other evidence favoring the prosecution might *566 well suffice to bar relief"); ante, at 552-553 ("If considered in isolation, a reasonable jury might well disregard [the evidence pointing to Mr. Muncey]. In combination, however, with the challenges to the blood evidence and the lack of motive with respect to House, the evidence pointing to Mr. Muncey likely would reinforce other doubts as to House's guilt"). According to the majority, House has picked the trifecta of evidence that places conviction outside the realm of choices any juror, acting reasonably, would make. Because the case against House remains substantially unaltered from the case presented to the jury, I disagree.

At trial, the State presented its story about what happened on the night of Mrs. Muncey's murder. The Munceys' daughter heard a deep-voiced perpetrator arrive at the Muncey home late at night and tell Mrs. Muncey that her husband had been in a wreck near the creek. App. 19. Ms. Tharp relayed her testimony again at the evidentiary hearing, and the District Court determined that she was a "very credible witness." Id., at 270, 323.

When police questioned House after witnesses reported seeing him emerge from the embankment near Mrs. Muncey's body shortly before it was discovered, he told two different officers that he never left Donna Turner's trailer the previous evening, even recounting the series of television programs he watched before going to bed. 7 Tr. 963-965, 1031-1032. He had worked to concoct an alibi we now know was a lie. On the day Mrs. Muncey's body was found, Bill Breeding, a criminal investigator at the Union County Sheriff's Office, observed House at the local jail and noticed that he had abrasions "across his knuckles and about his hands," two or three bruises on his right arm, scratches on his chest, and his right ring finger was red and swollen. 6 id., at 801-802. The interviewing officers noticed similar injuries. App. 78-80; 7 Tr. 974-975. House told them that his finger was swollen because he fell off a porch, and the scratches and bruises were from tearing down a building, and from a *567 cat. Ibid. Ms. Turner initially confirmed House's alibi, but she changed her story when police warned her that covering up a homicide was a serious offense. Id., at 1063. Ms. Turner then told police that House had in fact left her house that night between 10:30 and 10:45 p.m. Id., at 1062-1063. He came back some time later panting and sweating, shirtless and shoeless, and with various injuries. App. 88-91; 8 Tr. 1154-1155.

Also on the day the body was found, Sheriff Earl Loy asked House if he was wearing the same clothes he wore the night before. 6 id., at 845. House "hesitated," then stated that he had changed his shirt, but not his jeans. Ibid. In other words, he specifically tried to conceal from the police that he had worn other jeans the night before, for reasons that were to become clear. Ms. Turner revealed that House's statement that he had not changed his jeans was a lie, and police retrieved House's dirty jeans from Ms. Turner's hamper. Ibid. Of course, FBI testing revealed that House's jeans were stained with Mrs. Muncey's blood, and the District Court determined that House's new evidence of blood spillage did not undermine those test results. App. 348. If in fact Mrs. Muncey's blood only got on House's jeans from later evidentiary spillage, House would have had no reason to lie to try to keep the existence of the concealed jeans from the police.

Through Ms. Turner's testimony at trial, the jury also heard House's story about what happened that night. He left Ms. Turner's trailer late at night to go for a walk. Id., at 86. When he returned some time later—panting, sweating, and missing his shirt and shoes—he told her that some men in a truck tried to kill him. Id., at 88-91. When Ms. Turner asked House about his injuries, he attributed them to fighting with his assailants. Id., at 90; 8 Tr. 1154-1155. House retold this story to the District Court, saying that he initially lied to police because he was on parole and did not want to draw attention to himself. Record, Doc. 276, at 99, *568 108-109. In other words, having nothing to hide and facing a murder charge, House lied—and when he was caught in the lie, he said he lied not to escape the murder charge, but solely to avoid unexplained difficulties with his parole officer. The jury rejected House's story about the night's events, and the District Court "considered Mr. House's demeanor and found that he was not a credible witness." App. 329.

The jury also heard House's attempt to implicate Mr. Muncey in his wife's murder by calling Mrs. Muncey's brother, Ricky Green, as a witness. Mr. Green testified that two weeks before the murder, his sister called him to say that she and Mr. Muncey had been fighting, that she wanted to leave him, and that she was scared. 7 Tr. 1088. Mr. Green also testified that the Munceys had marital problems, and that he had previously seen Mr. Muncey hit his wife. Id., at 1087. The jury rejected House's attempt to implicate Mr. Muncey, and the District Court was not persuaded by House's attempt to supplement this evidence at the evidentiary hearing, finding that his new witnesses were not credible. App. 348.

Noticeably absent from the State's story about what happened to Mrs. Muncey on the night of her death was much mention of the semen found on Mrs. Muncey's clothing. House's single victory at the evidentiary hearing was new DNA evidence proving that the semen was deposited by Mr. Muncey. The majority identifies the semen evidence as "[c]entral to the State's case" against House, ante, at 528, but House's jury would probably be quite surprised by this characterization. At trial, Agent Bigbee testified that from the semen stains on Mrs. Muncey's clothing, he could determine that the man who deposited the semen had type A blood, and was a secretor. App. 54-56. Agent Bigbee also testified that House and Mr. Muncey both have type A blood, that House is a secretor, and that "[t]here is an eighty (80%) percent chance that [Mr. Muncey] is a secretor." Id., at 55-56; 6 Tr. 952 (emphasis added). Moreover, Agent Bigbee *569 informed the jury that because 40 percent of people have type A blood, and 80 percent of those people are secretors, the semen on Mrs. Muncey's clothing could have been deposited by roughly one out of every three males. Id., at 957. The jury was also informed several times by the defense that Mrs. Muncey's body was found fully clothed. See, e. g., 4 id., at 628; 9 id., at 1274.

The majority describes House's sexual motive as "a central theme in the State's narrative linking House to the crime," and states that without the semen evidence, "a jury . . . would have found it necessary to establish some different motive, or, if the same motive, an intent far more speculative." Ante, at 541. The State, however, consistently directed the jury's attention away from motive, and sexual motive was far from a "central theme" of the State's case— presumably because of the highly ambiguous nature of the semen evidence recounted above. The Tennessee Supreme Court did not mention that evidence in cataloging the "[p]articularly incriminating" or "[d]amaging" evidence against House. App. 135. The State did not mention the semen evidence in its opening statement to the jury, instead focusing on premeditation. 4 Tr. 613-615. The defense used its opening statement to expose lack of motive as a weakness in the State's case. Id., at 628. After the State's equivocal presentation of the semen evidence through Agent Bigbee's testimony at trial, the State again made no reference to the semen evidence or to a motive in its closing argument, prompting the defense to again highlight this omission. 9 id., at 1274 ("[W]hy was Carolyn Muncey killed? We don't know. Is it important to have some motive? In your minds? What motive did Paul Gregory House have to go over and kill a woman that he barely knew? Who was still dressed, still clad in her clothes").

In rebuttal, the State disclaimed any responsibility to prove motive, again shifting the jury's focus to premeditation:

*570 "The law says that if you take another person's life, you beat them, you strangle them, and then you don't succeed, and then you kill them by giving them multiple blows to the head, and one massive blow to the head, and that that causes their brains to crash against the other side of their skull, and caused such severe bleeding inside the skull itself, that you die—that it does not make any difference under God's heaven, what the motive was. That is what the law is. The law is that if motive is shown, it can be considered by the jury as evidence of guilt. But the law is that if you prove that a killing was done, beyond a reasonable doubt, by a person, and that he premeditated it, he planned it, it is not necessary for the jury to conclude why he did it." App. 106.

As a followup to this explanation, when the trial was almost over and only in response to the defense's consistent prodding, the State made its first and only reference to a possible motive, followed immediately by another disclaimer:

"Now, you may have an idea why he did it. The evidence at the scene which seemed to suggest that he was subjecting this lady to some kind of indignity, why would you get a lady out of her house, late at night, in her night clothes, under the trick that her husband has had a wreck down by the creek? . . . Why is it that you choke her? Why is it that you repeatedly beat her? Why is it that she has scrapes all over her body? Well, it is because either you don't want her to tell what indignities you have subjected her to, or she is unwilling and fights against you, against being subjected to those indignities. . . . That is what the evidence at the scene suggests about motive. But motive is not an element of the crime. It is something that you can consider, or ignore. Whatever you prefer. The issue is not motive. The issue is premeditation." Id., at 106-107.

*571 It is on this "obliqu[e]" reference to the semen evidence during the State's closing argument that the majority bases its assertion that House's sexual motive was a "central theme in the State's narrative." Ante, at 531, 541. Although it is possible that one or even some jurors might have entertained doubt about House's guilt absent the clearest evidence of motive, I do not find it more likely than not that every juror would have done so, and that is the legal standard under Schlup. The majority aphoristically states that "[w]hen identity is in question, motive is key." Ante, at 540. Not at all. Sometimes, when identity is in question, alibi is key. Here, House came up with one—and it fell apart, later admitted to be fabricated when his girlfriend would not lie to protect him. Scratches from a cat, indeed. Surely a reasonable juror would give the fact that an alibi had been made up and discredited significant weight. People facing a murder charge, who are innocent, do not make up a story out of concern that the truth might somehow disturb their parole officer. And people do not lie to the police about which jeans they were wearing the night of a murder, if they have no reason to believe the jeans would be stained with the blood shed by the victim in her last desperate struggle to live.

In Schlup, we made clear that the standard we adopted requires a "stronger showing than that needed to establish prejudice." 513 U.S., at 327. In other words, House must show more than just a "reasonable probability that . . . the factfinder would have had a reasonable doubt respecting guilt." Strickland v. Washington, 466 U.S. 668, 695 (1984). House must present such compelling evidence of innocence that it becomes more likely than not that no single juror, acting reasonably, would vote to convict him. Schlup, supra, at 329. The majority's conclusion is that given the sisters' testimony (if believed), and Dr. Blake's rebutted testimony about how to interpret Agent Bigbee's enzyme marker analysis summary (if accepted), combined with the revelation that the semen on Mrs. Muncey's clothing was deposited *572 by her husband (which the jurors knew was just as likely as the semen having been deposited by House), no reasonable juror would vote to convict House. Ante, at 553-554. Given the District Court's reliability findings about the first two pieces of evidence, the evidence before us now is not substantially different from that considered by House's jury. I therefore find it more likely than not that in light of this new evidence, at least one juror, acting reasonably, would vote to convict House. The evidence as a whole certainly does not establish that House is actually innocent of the crime of murdering Carolyn Muncey, and accordingly I dissent.

NOTES

[*] Briefs of amici curiae urging reversal were filed for the American Bar Association by Michael S. Greco, Rory K. Little, and Seth P. Waxman; for Former Prosecutors et al. by Andrew H. Schapiro, Timothy C. Lambert, George H. Kendall, and Miriam Gohara; and for the Innocence Project, Inc., by David Goldberg.

A brief of amici curiae urging affirmance was filed for the State of California et al. by Bill Lockyer, Attorney General of California, Manuel M. Medeiros, State Solicitor General, Robert R. Anderson, Chief Assistant Attorney General, Mary Jo Graves, Senior Assistant Attorney General, Patrick J. Whalen, Deputy Attorney General, and Ward A. Campbell, Supervising Deputy Attorney General, by Christopher L. Morano, Chief State's Attorney of Connecticut, and by the Attorneys General for their respective States as follows: Mike Beebe of Arkansas, John W. Suthers of Colorado, M. Jane Brady of Delaware, Charles J. Crist, Jr., of Florida, Lawrence G. Wasden of Idaho, Phill Kline of Kansas, Charles C. Foti, Jr., of Louisiana, Jim Hood of Mississippi, Mike McGrath of Montana, Thomas W. Corbett, Jr., of Pennsylvania, Lawrence E. Long of South Dakota, Greg Abbott of Texas, and Rob McKenna of Washington.