Respondent Mike Mullin, warden of the Oklahoma State Penitentiary, appeals the district court's decision granting conditional habeas relief in the form of a new sentencing hearing to petitioner Sharieff Sallahdin, an Oklahoma state prisoner convicted of first degree murder and sentenced to death. We exercise jurisdiction pursuant to 28 U.S.C. § 1291 and reverse.

The relevant underlying facts of Sallahdin's crime and state court proceedings were set forth in Sallahdin v. Gibson, 275 F.3d 1211, 1221 (10th Cir.2002) (Sallahdin I ):

At approximately 5:00 a.m. on October 21, 1991, James Principe and Bradley Grooms were stocking shelves at the 7-Eleven convenience store where they worked [in Lawton, Oklahoma]. Principe heard a loud bang and saw a black man looking in Grooms' direction. Principe ducked, made his way to the back of the store and locked himself in the bathroom. After emerging from the bathroom, he contacted the police and then saw Grooms lying motionless on the floor with a gunshot wound to the chest. Grooms died from his injuries. Principe later identified Sallahdin as the man who shot Grooms.

That same morning, Lynn Smith stopped at the 7-Eleven to get some ice. Sallahdin was behind the counter and gave her a cup of ice. She did not see anyone else in the store. Upon leaving, she looked back and saw Sallahdin leave the store and drive away in a car. The following day, Sallahdin was taken into custody at his wife's home in Akron, Ohio.

At trial, Sallahdin testified another man committed the murder. The jury convicted Sallahdin of first degree malice murder, which is punishable by death in Oklahoma. When the trial proceeded to the sentencing phase, the State sought the death penalty based on three aggravators: (1) Sallahdin posed a continuing threat to society; (2) he committed the murder to avoid arrest or prosecution; and (3) he knowingly created a great risk of death to more than one person. In addition to the guilt phase evidence about the crime, the State presented evidence of threats Sallahdin made while incarcerated. The jury found all three aggravating circumstances. After Sallahdin presented his own testimony in mitigation and mitigating testimony from friends, family, commanders, peers and others who knew him; the jury determined the aggravating factors outweighed the mitigating evidence and fixed punishment at death.

On direct appeal, the Oklahoma Court of Criminal Appeals affirmed the conviction and death sentence, after striking the great risk of death to more than one person aggravator and reweighing the remaining aggravators against the mitigating evidence. That court later denied post-conviction relief. Sallahdin v. State, 947 P.2d 559 (Okla.Crim.App.1997).

In Sallahdin I, we affirmed the denial of all of Sallahdin's federal habeas issues except one. We found "troubling" Sallahdin's assertion that his trial counsel was ineffective for failing to present second-stage mitigating evidence of the effects of Sallahdin's steroid use on his behavior at the time of the crime. Id. at 1220. With respect to that issue, we concluded:

After carefully examining the record, we are persuaded that, had this evidence been presented, there is a reasonable probability the outcome of the sentencing phase would have been different, i.e., that the jury would have imposed a sentence other than death. Because, however, we are not privy to trial counsel's reasons for not presenting this evidence, and because we can envision circumstances where it would have been constitutionally reasonable for counsel not to introduce this evidence despite its potentially mitigative effect, we find it necessary to reverse and remand this case to the district court for an evidentiary hearing solely on this specific issue concerning trial counsel's performance.

Id. With respect to the evidentiary hearing on remand, we stated:

[T]he purpose of the evidentiary hearing will be to determine trial counsel's reasons, or lack thereof, for foregoing the use of [defense expert] Dr. Pope's testimony during the sentencing phase. If trial counsel made a strategic decision not to use Dr. Pope's testimony, the district court will then need to assess whether that was a constitutionally reasonable decision under the circumstances. If, however, it is established that trial counsel was neglectful, or otherwise erred, in failing to call Dr. Pope as a second-stage witness, then trial counsel's performance cannot be deemed constitutionally reasonable. In turn, Sallahdin would be entitled to federal habeas relief in the form of a new sentencing proceeding.

Id. at 1240-41.

During the evidentiary hearing on remand, Sallahdin maintained that his trial counsel, Mark Barrett, was neglectful in failing to present evidence of his steroid use. Sallahdin relied exclusively on the testimony of Barrett to support the theory. Barrett, who appeared primarily through videotaped deposition testimony, testified that at some point during his two-year representation of Sallahdin he hired Dr. Harrison Pope, a psychiatrist and Harvard professor, as an expert on the psychological effects of steroids. According to Barrett, Pope examined Sallahdin and concluded that Sallahdin "had suffered extreme negative effects from the use of steroids, and, particularly important in this case, from withdrawal of the use of steroids." Barrett Dep. at 10. Barrett testified that, prior to trial, he and Sallahdin were in agreement that Pope's testimony would be helpful to the defense and should be presented during the second-stage proceedings (assuming Sallahdin was found guilty at the conclusion of the first-stage proceedings). Approximately four days prior to trial, Barrett submitted a written request to the director of the Oklahoma Indigent Defense System (OIDS) requesting funding for Pope's appearance at trial and the request was approved by the director of OIDS on the day it was submitted. According to Barrett, he had some recollection that, on the evening after the jury returned its first-stage verdict and prior to the beginning of the second-stage proceedings, he asked another attorney in his office "to look into getting Dr. Pope there." Id. at 26. Barrett conceded, however, that Pope ultimately did not appear as a second-stage witness, but he was unable to recall why he did not call Pope as a witness — he did not think he made a strategic decision to not call Pope, nor did he think not calling Pope was the result of his negligence. When asked if he made a strategic decision not to present Pope's testimony, Barrett testified: "I can't imagine that I did. I mean the difficulty in answering that with any certainty is because there's no explanation for all this that makes any sense to me. But it was important enough evidence to us that I can't imagine making any such [strategic] decision." Id. at 27. Barrett testified that he did not remember seeking Sallahdin's approval to forego the presentation of Pope as a second-stage witness. Barrett testified that he vaguely remembered approaching the bench during the second-stage proceedings and making an offer of proof as to Pope's testimony, but conceded that the transcript of the state court proceedings contained no such bench conference or offer of proof. When asked if he simply neglected to present Pope's testimony, Barrett testified he did not think that was a "strong ... possibility," but admitted that "everything [wa]s a possibility." Id. at 35. On cross-examination, Barrett again admitted that neglect was "a possibility," but questioned whether that was in fact what had occurred: "I mean, it's not something I could have just overlooked, hey, you know, the trial was going on, I just kind of forgot about Dr. Pope. I mean, he's a Harvard professor who was probably the world's leading expert on steroid use. The fact that he was out there wouldn't have just slipped my mind." Id. at 58. Thus, Barrett carefully walked the line between admitting that not calling Pope as a witness was a strategic decision and admitting it was the result of his negligence.

Although respondent did not have the burden of proof, he presented evidence from four witnesses to counter Sallahdin's theory that counsel was negligent. Dr. John Call, a forensic psychologist hired by the prosecution, testified that he examined Sallahdin approximately two months before trial and, during the examination, Sallahdin admitted he had not used steroids since April 1991 (approximately five to six months prior to the homicide). Call further testified that on August 23, 1993, approximately two weeks prior to trial, he prepared and sent to the prosecutors a memo stating, in pertinent part, that "by his own admission, Mr. [Sallahdin] had not been using steroids for many months prior to" the homicide. Evidentiary hearing tr. at 71.

Fred Smith, Jr., the assistant district attorney who helped prepare the Sallahdin case for trial, testified that he received and was aware of the contents of Call's memo. More importantly, Smith testified that he recalled pretrial discussions with Barrett in which Barrett stated he was not going to present steroid-related evidence "as a mitigator in the second phase of the trial." Id. at 49. According to Smith, Barrett stated his decision to forego the defense was based on "the defendant having stopped using steroids some time before the murder itself" and because he "didn't want to run the risk of antagonizing or alienating the jury by presenting this type of argument." Id. at 49-50.

Robert Schulte, the district attorney who tried the case against Sallahdin, testified that, in accordance with office policy, Call's memo would have been hand-delivered to Sallahdin's counsel (but admitted he had no independent recollection of giving the memo to Barrett). Schulte testified that Smith informed him prior to trial that steroids would not be part of the trial, and that steroids were never an issue at trial. To counter Barrett's suggestion that the trial judge may have prohibited the defense from presenting any evidence pertaining to steroid use, Schulte testified that it was the trial judge's standard practice to make a record of everything.

The final witness for the respondent was Darrell Dawkins, a detective who worked on the Sallahdin case. Dawkins testified that he participated in a post-arrest interview of Sallahdin (approximately seven or eight days after the homicide) in which Sallahdin indicated it had been "six months or longer" since he had used steroids. Id. at 113. Dawkins acknowledged, however, that he did not take any notes during the interview and had not included this information in his report of the interview since, according to him, he did not know that steroid use would be an issue at trial.

Sallahdin presented rebuttal evidence from two witnesses. Dale Johnson, an acquaintance from Sallahdin's hometown of Akron, Ohio, testified that he and Sallahdin had lifted weights together for many years, and that he and Sallahdin had regularly purchased and used steroids. Johnson further testified that Sallahdin's behavior was radically different when he was using steroids — when using steroids, he was aggressive and reckless; when not using steroids, he was quiet and reserved. According to Johnson, Sallahdin traveled to Akron in August or September 1991 (approximately a month or two prior to the homicide) and, during the visit, purchased approximately $250.00 worth of steroids.

Sallahdin's second rebuttal witness was Barrett, who denied telling prosecutors prior to the first-stage verdict that he would not be calling Pope as a witness in the second stage. Barrett testified that he would not have prepared and submitted a request for funding for Pope's trial appearance if he had informed the prosecution he would not be presenting Pope's testimony. Barrett further testified that Call's memo would not have affected his decision because the defense had evidence (presumably Johnson's testimony) that Sallahdin had used steroids in Akron shortly before the homicide. Barrett also testified that, in light of how quickly the jury returned its first-stage verdict, it was not his plan to rely on a residual doubt theory during the second stage. On cross-examination, Barrett admitted that he had no independent memory of what transpired and was testifying based on what he thought would have been consistent with the circumstances. Barrett also admitted that he was certain he did not have any conversations with Pope after the trial began.

On March 24, 2003, the district court granted Sallahdin conditional habeas relief. The court examined Barrett's testimony and rejected his suggestion that the trial court may have prohibited him from presenting Pope's testimony. ROA, Doc. 77 at 3 ("The trial record reveals no such ruling by the court."). The court further concluded, based upon Barrett's testimony, there was "no evidence ... that trial strategy was the basis for failing to proffer Dr. Pope's testimony as a witness during sentencing." Id. The court acknowledged the respondent's contention "that the likely reason ... Barrett did not offer Dr. Pope's testimony was the discovery that [Sallahdin] had ceased taking steroids months before the murder." Id. The court rejected this contention, however, on the basis of Barrett's testimony that he had independent evidence that Sallahdin "had taken steroids near the time of the crime," that he had no recollection of Call's memo, and that he had no recollection of being informed by the police "that his client had stopped his steroid usage before the murder." Id. at 4. Ultimately, the district court concluded: Th[e] evidence establishes that there is no discernible explanation for counsel's failure to call Dr. Pope to the stand on Petitioner's behalf. Even applying the deferential standard mandated by Strickland[ v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984)], the Court cannot conclude that Mr. Barrett's decision not to present the testimony was the result of trial strategy given the importance of the steroid testimony and the absolute lack of any basis for failing to present it. Fees and expenses had been approved for Dr. Pope and Dr. Pope knew that he would be needed with little advance notice should Petitioner be convicted. Despite being able to recall this information Mr. Barrett is unable to remember why Dr. Pope did not appear and why no proffer of evidence was made. Nothing in the trial court record indicates that the court excluded the evidence or that court-imposed or other time constraints prevented Mr. Pope from appearing at the trial. Petitioner has met his burden under the first prong of Strickland.

Id. The court ordered the State of Oklahoma "to commence a new sentencing proceeding" within 180 days from the date of its order. Id. at 5. The court stated that "[f]ailure to commence such a proceeding," would "result in ... vacation of [Sallahdin]'s death sentence and an order sentencing [Sallahdin] to life imprisonment." Id. The district court stayed its conditional grant of habeas relief pending appeal.

Respondent on appeal challenges both the district court's factual findings and its legal conclusions regarding the first prong of Sallahdin's ineffective assistance claim. Reviewing the district court's findings of fact for clear error and its conclusions of law de novo, see Sallahdin I, 275 F.3d at 1222, we agree with respondent that the district court erred, both factually and legally, in resolving Sallahdin's claim. In particular, we conclude the district court failed to properly consider and apply the evidentiary burden that accompanies a Strickland claim, failed to give proper consideration to evidence presented at the evidentiary hearing and contained in state court records, and erred in concluding there could have been no reasonable strategic basis for foregoing the steroid-related evidence during the second-stage proceedings.

To prevail on the first prong of the Strickland test, a criminal defendant such as Sallahdin bears the burden of establishing that his trial counsel "made errors so serious that counsel was not functioning as the `counsel' guaranteed the defendant by the Sixth Amendment." Strickland v. Washington, 466 U.S. 668, 687, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984). Review of counsel's performance under this prong is "highly deferential." Id. at 689, 104 S. Ct. 2052. "[A] court must indulge a strong presumption that counsel's conduct falls within the wide range of reasonable professional assistance; that is, the defendant must overcome the presumption that, under the circumstances, the challenged action might be considered sound trial strategy." Id. (internal quotations omitted); see Yarborough v. Gentry, 540 U.S. 1, 124 S. Ct. 1, 5, 157 L. Ed. 2d 1 (2003) ("When counsel focuses on some issues to the exclusion of others, there is a strong presumption that he did so for tactical reasons rather than through sheer neglect."); Kimmelman v. Morrison, 477 U.S. 365, 384, 106 S. Ct. 2574, 91 L. Ed. 2d 305 (1986) ("Counsel's competence... is presumed."); Strickland, 466 U.S. at 690, 104 S. Ct. 2052 (stating "counsel is strongly presumed to have rendered adequate assistance and made all significant decisions in the exercise of reasonable professional judgment"). To surmount the strong presumption of reasonable professional assistance, a criminal defendant bears the burden of proving, by a preponderance of the evidence, that his trial counsel acted unreasonably. See Kimmelman, 477 U.S. at 384, 106 S. Ct. 2574; Alcala v. Woodford, 334 F.3d 862, 869 (9th Cir.2003); Putman v. Head, 268 F.3d 1223, 1243 (11th Cir.2001); Cody v. United States, 249 F.3d 47, 52 (1st Cir.2001); see also Strickland, 466 U.S. at 688, 104 S. Ct. 2052 (holding, without reference to any specific evidentiary standard, that a criminal defendant "must show that counsel's representation fell below an objective standard of reasonableness"); Wayne R. LaFave et al., Criminal Procedure § 11.10(c) at 715 (West 2d 1999) (noting Strickland"places upon the defendant the burden of showing that counsel's action or inaction was not based on a valid strategic choice").

These principles play a critical role in this case. As applied to the particular facts of this case, it was Sallahdin's burden on remand to overcome the strong presumption of constitutionally reasonable conduct and prove, by a preponderance of the evidence, that trial counsel acted negligently in failing to present steroid-related evidence, including the testimony of Pope, during the second-stage proceedings. Significantly, this "burden of persuasion on the constitutional issue of competence" remained continually on Sallahdin. Chandler v. United States, 218 F.3d 1305, 1314 n. 15 (11th Cir.2000) (en banc). "Never d[id] the government [on remand] acquire the burden to show competence, even when some evidence to the contrary might [have] be[en] offered by" Sallahdin. Id.

With these principles in mind, we turn to the district court's findings and conclusions. The court's initial statement that the "evidence establishes that there is no discernible explanation for counsel's failure to call Dr. Pope" is telling. ROA, Doc. 77 at 4. This statement confirms that Sallahdin had not carried his burden and most certainly had not overcome the strong presumption that counsel's conduct fell within the wide range of reasonable professional assistance. Rather, the district court, by applying a process of elimination, implicitly found that trial counsel was neglectful in failing to present steroid-related evidence during the second-stage proceedings. In particular, the court noted there was no evidence that the trial court "excluded the evidence" or "that court-imposed or other time constraints prevented Mr. Pope from appearing at the trial." Id. Further, the court rejected the possibility that trial counsel made a strategic decision not to present steroid-related evidence: "[T]he Court cannot conclude that Mr. Barrett's decision not to present the testimony was the result of trial strategy given the importance of the steroid testimony and the absolute lack of any basis for failing to present it." Id.

After examining the record on appeal, we agree with the district court on three points. First, we agree there is no evidence in the record indicating that the trial court prohibited Sallahdin's trial counsel from introducing steroid-related evidence during the second-stage proceedings, and there is otherwise no evidence in the record to support such a finding. Although Barrett testified that he recalled having several off-the-record discussions with the trial court and the prosecutors, he did not testify that he was prohibited by the trial court from introducing steroid-related evidence during the second-stage proceedings. Second, we agree there was no evidence indicating the time constraints imposed by the trial schedule prevented Pope from appearing at the second-stage proceedings. Barrett did testify to having some recollection of asking another attorney in his office "to look into getting Dr. Pope there" after the conclusion of the first-stage proceedings. Barrett Dep. at 26. However, this testimony is not sufficient, by itself, to allow an inference that Barrett and his staff actually attempted, but were unable for some reason, to obtain Pope's presence at trial the following day.1 Third, because it was uncontroverted that fees and expenses for Pope's appearance at trial were approved by the director of OIDS prior to trial, there was no evidence that budgetary constraints prevented Pope from appearing at the second-stage proceedings.

That leaves only two possible explanations for trial counsel not presenting any steroid-related evidence during the second-stage proceedings. The first possibility is that trial counsel made a strategic decision not to present such evidence. The second possibility is that trial counsel simply forgot to present such evidence. The district court found that the latter occurred. The district court concluded that Sallahdin had overcome the strong presumption of reasonable professional conduct outlined in Strickland and was entitled to federal habeas relief on the basis of his ineffective assistance of counsel claim.

We reject the district court's legal conclusions and its factual finding that trial counsel was negligent for several reasons. The district court's first, and perhaps most critical, misstep arose in failing to hold Sallahdin to his burden of proof under Strickland when reviewing Barrett's testimony. During the evidentiary hearing, Barrett was asked whether he made a strategic decision to forego Pope's testimony or instead was neglectful. In response, he expressed equal skepticism regarding both possibilities. Barrett Dep. at 27 (when asked if he made a strategic decision not to call Pope, Barrett responded: "[I]t was important enough evidence to us that I can't imagine making any such decision"), 58 (when asked about the possibility of neglect, Barrett indicated "that [did not] seem likely," and stated: "I mean, he's a Harvard professor who was probably the world's leading expert on steroid use. The fact that he was out there wouldn't have just slipped my mind."). Indeed, Barrett admitted he had no independent memory of what transpired and that he was testifying based upon what he thought would have been consistent with the surrounding circumstances. Thus, while the district court was correct in noting that Barrett's testimony provided "no discernible explanation for [his] failure to call Dr. Pope to the stand on Petitioner's behalf," ROA, Doc. 77 at 4, it was equally true that his testimony provided no basis for finding he was neglectful. In other words, Barrett's testimony, such as it was, was insufficient to surmount the presumption that the failure to present steroid-related evidence was the product of reasonable professional judgment. See Williams v. Head, 185 F.3d 1223, 1228 (11th Cir.1999) (presuming, where petitioner's trial counsel was unable to recall many of his thought processes during trial, that counsel "did what he should have done, and that he exercised reasonable professional judgment"); see also Yarborough, 124 S.Ct. at 5-6 (indicating presumption of reasonable effective assistance is controlling in circumstances where reviewing court has "`no way of knowing whether a seemingly unusual or misguided action by counsel had a sound strategic motive'") (quoting Massaro v. United States, 538 U.S. 500, 505, 123 S. Ct. 1690, 155 L. Ed. 2d 714 (2003)).

The district court also, in our view, improperly discounted the testimony of the two prosecutors involved in the state trial proceedings. As noted, both witnesses (Smith and Schulte) testified that Barrett was provided with a copy of Call's memo prior to trial, and Smith testified that Barrett told him prior to trial he was going to forego the steroid-related evidence due, in part, to Sallahdin's admission to Call that he had not used steroids for several months prior to the homicide. In its order granting habeas relief, the district court noted that Barrett "denied any recollection" of the contents of Call's memo.2 ROA, Doc. 77 at 4. Given Barrett's faulty memory on the general subject of the steroid-related evidence, we fail to see how the district court reasonably could have given credence to his testimony regarding Call's memo. More specifically, we fail to see how Barrett's inability to remember could outweigh the testimony of the two prosecution witnesses who remembered the events at issue.3

Further, there is no indication that the district court took into account the records of the state trial and post-conviction proceedings which, in our view, are more consistent with a finding that Barrett made a strategic decision to forego steroid-related evidence during the second-stage proceedings than with a finding that Barrett simply forgot to present such evidence. The state post-conviction record indicates that, aside from Pope, Barrett had available at least four lay witnesses who could have testified about Sallahdin's steroid use and its apparent impact on his personality — his wife, his mother, his mother-in-law, and his friend Dale Johnson. The state trial record indicates that only his mother and Johnson actually testified during the second-stage proceedings and neither was questioned by Barrett on the subject of steroids. Indeed, the subject of steroids was not mentioned during second-stage proceedings. Instead, Barrett's questioning of most of the second-stage defense witnesses focused on Sallahdin's temperament, which was generally described as quiet, loving, kind, and nonviolent. In addition, the state trial record indicates that, immediately prior to the beginning of the second-stage proceedings, Barrett filed a set of proposed second-stage jury instructions. One of the proposed instructions listed various mitigating circumstances, including the following: "[Sallahdin's] state of mind was adversely affected by the effect of steroids." State Trial Record, Vol. 4 at 157. At the conclusion of the second-stage evidence, the trial judge modified the tendered mitigating circumstances instruction and excised the quoted reference to steroids. Significantly, there is no indication in the state court trial record that Barrett objected to the modification. In sum, this evidence suggests that Barrett intentionally decided not to rely on Sallahdin's steroid use as a mitigating factor during the second-stage proceedings.

Lastly, it appears the district court's factual finding of neglect was impacted by its erroneous conclusion that there could have been no strategic "basis for failing to present" the steroid-related evidence. ROA, Doc. 77 at 4. In our view, the record on appeal supplies at least three reasons why "it would have been constitutionally reasonable for counsel not to introduce this evidence despite its potentially mitigative effect." Sallahdin I, 275 F.3d at 1220. First, the jury could have viewed Sallahdin's steroid use in a negative light. Not only are anabolic steroids illegal, the jury could have drawn negative inferences from the fact that Sallahdin knowingly used the substances for a period of approximately four years, during which time he allegedly exhibited various signs of aggressiveness and hostility toward friends and family. In other words, the jury reasonably might have concluded that Sallahdin's steroid use was reckless and that Sallahdin knew or should have known that it might result in negative consequences. Second, there was a significant risk that the jury, having found Sallahdin to be less than a credible witness during the first-stage proceedings, likewise would have discounted his steroid theory since a critical component of that theory (Pope's scientific conclusions) rested heavily on Sallahdin's self-reports of his psychiatric symptoms. Third, the evidence presented by respondent during the evidentiary hearing on remand indicates the prosecution was in possession of a powerful rebuttal to Sallahdin's steroid theory — post-arrest statements by Sallahdin indicating he had not in fact used steroids in close proximity to the murder. For any or all of these reasons, we conclude that Barrett reasonably could have decided not to present a steroid-related defense during the second-stage proceedings.4

Ultimately, given the strong presumption of reasonable professional assistance and the accompanying evidentiary burden imposed on Sallahdin, we agree with respondent that it is Sallahdin, rather than respondent, who must bear the brunt of Barrett's faulty memory and his ambiguous testimony. See Yarborough, 124 S.Ct. at 5 (suggesting such a result where evidence is insufficient to overcome presumption of reasonable professional assistance); Williams, 185 F.3d at 1227-28 (reaching similar conclusion where habeas petitioner's trial counsel unable to recall his thought processes during trial); Fretwell v. Norris, 133 F.3d 621, 623-24 (8th Cir.1998) (rejecting district court's reliance on trial counsel's inability to recall reasons underlying his performance as basis for establishing lack of competent performance); cf. United States v. Corrado, 304 F.3d 593, 605 (6th Cir.2002) ("The party with the burden, in effect, bears the risks of [witness] amnesia and faded memory because of the passage of time.").5 As the Eleventh Circuit noted in Chandler,"[a]n ambiguous or silent record is not sufficient to disprove the strong and continuing presumption" that counsel's performance was reasonable and that counsel made all significant decisions in the exercise of reasonable professional judgment. 218 F.3d at 1314 n. 15. Thus, we agree with respondent that the scales tipped in his favor at the start of the evidentiary hearing (i.e., in favor of the "strong presumption" that Barrett acted in a constitutionally reasonable fashion), and, given Barrett's inability to remember, remained in respondent's favor at the conclusion of the evidentiary hearing. We therefore conclude the district court erred in holding that Sallahdin satisfied the first prong of the Strickland analysis, and in turn granting Sallahdin conditional habeas relief in the form of a new sentencing proceeding.6

The judgment of the district court granting petitioner conditional habeas relief is REVERSED.

EBEL, Circuit Judge, dissenting.

Prior to the events giving rise to this case, Petitioner Sharieff Imani Sallahdin appears to have been a law-abiding man who had enjoyed a normal upbringing, had achieved some academic and athletic success, and was generally regarded as a good soldier. He also was a serious body-builder who, as he became ever more engrossed in his weight-lifting regimen, began to use steroids heavily. In the pre-dawn hours of October 21, 1991, Sallahdin entered a convenience store in Lawton, Oklahoma and shot and killed a clerk while attempting to rob the store. In general terms, the issue before us centers on whether Sallahdin's counsel was constitutionally ineffective in failing to call Dr. Pope as a witness in the penalty phase of the trial to testify that Sallahdin's previous steroid use could have explained the uncharacteristic act of violence resulting in the convenience store shooting.

In our previous ruling in this case, this panel unanimously held that "there is a reasonable probability that the presentation of Dr. Pope's testimony [explaining how Sallahdin's steroid use could have triggered his apparent transformation from non-violent and law-abiding citizen to cold-blooded murderer], could have altered the outcome of the sentencing phase." Sallahdin v. Gibson (Sallahdin I), 275 F.3d 1211, 1240 (10th Cir.2002). Although we remanded for an "evidentiary hearing on the issue of trial counsel's performance" id., I believe our prior opinion concluded that the prejudicial effect of counsel's refusal to call Dr. Pope would lead to a conclusion that counsel's performance was constitutionally defective unless the more deferential standard utilized to evaluate actual strategic choices by counsel was employed. Thus, in our remand we instructed the district court that, "If trial counsel made a strategic decision not to use Dr. Pope's testimony, the district court will then need to assess whether that was a constitutionally reasonable decision under the circumstances. If, however, it is established that trial counsel was neglectful, or otherwise erred, in failing to call Dr. Pope as a second stage witness, then trial counsel's performance cannot be deemed constitutionally reasonable [and] Sallahdin would be entitled to federal habeas relief..." Id. (Emphasis added.) The district court held such a hearing, and concluded that there was no evidence in the record that Sallahdin's counsel made a conscious strategic choice not to call Dr. Pope, and that counsel's performance was constitutionally deficient. Notwithstanding these findings by the district court, a majority of our panel now reverses the grant of habeas relief. I must respectfully dissent.

As the majority correctly notes, Strickland v. Washington instructs us to "indulge a strong presumption that counsel's conduct falls within the wide range of reasonable professional assistance." 466 U.S. at 689, 104 S. Ct. 2052. While this presumption can be rebutted by the defendant's showing "that counsel's representation fell below an objective standard of reasonableness," Id. at 688, 104 S. Ct. 2052, the bar is set very high. An attorney's performance will not fall short of this baseline if we conclude his or her actions are consistent with any possible strategy that would have been objectively sound under the circumstances of the case.

It is critical to remember, however, that our focus at this point is on an objective analysis, and not on subjective thought processes. Thus we held in Bullock v. Carver that defense counsel's failure to object to hearsay evidence was not objectively unreasonable, even though the defendant's counsel subjectively was unaware that the evidence could have been excluded under state law, because a hypothetical "fully informed attorney could have concluded that admitting the hearsay statement was to [the defendant's] strategic advantage." 297 F.3d 1036, 1053-54 (10th Cir.2002). That the particular attorney in that case never followed that line of thought was immaterial to our conclusion that his actions had been objectively reasonable. Similarly, in Bonin v. Calderon, 59 F.3d 815, 838 (9th Cir.1995), the Ninth Circuit rejected an ineffective assistance claim brought by a defendant who claimed his counsel had abused drugs during trial, finding the accusation to be irrelevant where the attorney's performance at trial had not been objectively unreasonable. See also Chandler v. United States, 218 F.3d 1305, 1315 and n. 16 (11th Cir.2000) (en banc) ("[T]o show that the conduct was unreasonable, a petitioner must establish that no competent counsel would have taken the action that his counsel did take.... We look at the acts or omissions of counsel that the petitioner alleges are unreasonable and ask whether some reasonable lawyer could have conducted the trial in that manner. Because the standard is an objective one, that trial counsel ... admits that his performance was deficient matters little.")

A successful showing by the defendant that no reasonable strategy can explain or justify his lawyer's conduct does not, however, end our inquiry in all cases. When counsel seeks to justify his or her behavior on the basis of an actual subjective strategy, we give an extra bump to the presumption of reasonableness. That is, if the lawyer in fact pursued an actual informed legal strategy, it will be even more likely that we will find the approach to be objectively reasonable than if the approach taken was random or reflexive conduct without strategic thought behind it. Bullock v. Carver, 297 F.3d at 1046; Moore v. Marr, 254 F.3d 1235, 1239 (10th Cir.2001).

As we stated recently in our en banc decision in Bryan v. Mullin,"although the ultimate question is always whether counsel's performance fell below an objective standard of reasonableness, `where it is shown that a challenged action was, in fact, an adequately informed strategic choice, we heighten our presumption of objective reasonableness and presume that the attorney's decision is nearly unchallengeable.' "335 F.3d 1207, 1223 n. 23 (quoting Bullock v. Carver, 297 F.3d at 1047, 1051). An attorney who is shown to have made a conscious strategic choice will only be found constitutionally incompetent where the strategy pursued was "completely unreasonable, not merely wrong." Bullock, 297 F.3d at 1047. In Bullock, we summarized the legal framework of the interaction between objective reasonableness and subjective strategy as follows:

[T]he overriding question under the first prong of Strickland is whether, under all the circumstances, counsel performed in an objectively unreasonable manner. Two presumptions inform our objective reasonableness inquiry. First, we always start the analysis [with the presumption] that an attorney acted in an objectively reasonable manner and that an attorney's challenged conduct might have been part of a sound trial strategy. Second, where it is shown that a particular decision was, in fact, an adequately informed strategic choice, the presumption that the attorney's decision was objectively reasonable becomes "virtually unchallengeable." However, it is important to remember that these presumptions are simply tools that assist us in analyzing Strickland's deficient performance prong and they do not, in and of themselves, answer the ultimate question, which is whether counsel performed in an objectively reasonable manner. So, for example, even though counsel's strategy was ill-informed and thus does not qualify for the virtually unchallengeable presumption of reasonableness, a court reviewing the record before it might still conclude that counsel performed in an objectively reasonable manner. And, conversely, it is also possible on rare occasions to conclude that counsel's fully-informed strategic choices were unreasonable if "`the choice was so patently unreasonable that no competent attorney would have made it.'" Phoenix v. Matesanz, 233 F.3d 77, 82 n. 2 (1st Cir.2000) (quoting Washington v. Strickland, 693 F.2d 1243, 1254 (5th Cir.1982)).

297 F.3d at 1046.

Whether or not counsel actually relied on an identified strategy is a factual question, not predicated on any legal presumptions about hypothetical or potential strategies. Bryan, 335 F.3d at 1221 n. 17. Only if the evidence reveals that counsel was in fact pursuing an informed strategic approach do we then apply the even higher presumption such that the reasonableness of counsel's performance becomes "virtually unchallengeable" unless it is "completely unreasonable." If the record does not establish an informed strategy behind the challenged action, then the court must rely on the normal Strickland presumption of competent counsel, which can be rebutted by the defendant showing that there is no reasonable basis for taking that particular action.

Our remand made it clear that if Sallahdin's counsel's decision not to call Dr. Pope was evaluated without the support of an actual informed strategy, "then counsel's performance cannot be deemed constitutionally reasonable." But if the decision not to call Dr. Pope was the result of an actual informed strategy, then the extra bump that we give to the presumption of reasonableness might allow the district court to conclude that the decision was constitutionally sufficient. Thus, it became critical on remand for the district court to determine whether the decision not to call Dr. Pope was an informed strategic decision or not. We stated, "The purpose of the evidentiary hearing will be to determine trial counsel's reasons, or lack thereof, for foregoing the use of Dr. Pope's testimony during the sentencing phase."

The district court on remand concluded that there was no evidence that the decision not to call Dr. Pope was the result of an informed strategy. Unless that finding is to be set aside as clearly erroneous, we are left with the ordinary Strickland presumption of competence to apply. I believe our prior opinion has already determined, as the rule of the case, that such a presumption of competence has been rebutted on this issue. If, on the other hand, the majority is correct that that issue was not decided in our prior opinion, I would now decide it, contrary to the majority opinion, by holding that the decision not to call Dr. Pope was unreasonable and constituted ineffective assistance of counsel.

At the close of the evidentiary hearing on remand, the district court ruled out several possible reasons that might have motivated Barrett's choice not to call Dr. Pope. According to the district court, the state trial record indicated that Barrett was not prevented from calling Dr. Pope during the penalty phase by any court order excluding Dr. Pope's testimony. Nor did the state court record support the theory that Dr. Pope's absence was the result of financial or time constraints; Barrett had received authorization for Dr. Pope's expenses and fees four days before the start of Sallahdin's trial, and Dr. Pope was aware that he might be called on short notice. Finally, the district court found that the evidence presented on remand also ruled out the government's preferred theory — that Barrett had decided to abandon all mention of steroids at sentencing after being confronted with evidence that Sallahdin might have ceased his steroid use several months before committing the murder.

The district court's ultimate factual conclusion was stated quite plainly: "There has been no evidence presented that supports the conclusion that trial strategy was the basis for failing to proffer Dr. Pope[ ] as a witness during sentencing.... This evidence establishes that there is no discernible explanation for counsel's failure to call Dr. Pope to the stand on Petitioner's behalf."

"The question of whether an attorney's actions were actually the product of a tactical or strategic decision is an issue of fact." Bryan v. Mullin, 335 F.3d 1207, 1221 n. 17 (10th Cir.2003) (en banc). Although we may disagree with the district court's factual conclusions, we may overrule them only if we find the district court committed clear error. Id. at 1216; Rogers v. Gibson, 173 F.3d 1278, 1282 (10th Cir.1999). We may find clear error only if our review of all the evidence leaves us with a "definite and firm conviction that a mistake has been committed." Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S. Ct. 1504, 84 L. Ed. 2d 518 (1985). Our deference for the conclusions of the district court is especially heightened where the lower court's judgment rests on an evaluation of witnesses' credibility. Id. at 575, 105 S. Ct. 1504.

Barrett's testimony below was frequently vague and often marred by his failing memory. Where, however, the government's completely circumstantial evidence that Barrett had made a conscious strategic decision not to present Dr. Pope's testimony was countered by Barrett's specific disavowals as well as documentary evidence in the record, we simply cannot fairly say that the district court's fact finding on this point was clearly erroneous.

Thus, the finding on remand that there was no showing of an actual informed strategy not to call Dr. Pope must stand. Therefore, we are directed back to the ordinary Strickland rebuttable presumption of competence.

As I have stated previously, I believe this panel's unanimous opinion in Sallahdin I has already settled the objective reasonableness issue, and law of the case counsel against altering our judgment now. However, even if we were to revisit that previous holding, I would again conclude that Barrett's failure to put Dr. Pope on the stand was objectively unreasonable.

Although our prior opinion in Sallahdin I does speculate that "we cannot say that presentation of a steroid-use defense was without risk of negative consequences, or was the only reasonable second-stage strategy that trial counsel could have adopted," Sallahdin I, 275 F.3d at 1240, and footnote 10 specifically mentioned a "residual doubt" strategy as a possibility that might have been largely consistent with the mitigation evidence actually presented in the sentencing proceeding, Id. at 1240 n. 10, I believe this discussion was only in the context of leaving open on remand the possibility that if the district court found that Barrett had relied on a specific conscious strategy, it might then conclude that such a strategy was not "completely unreasonable."1 Our mention of "residual doubt" as a potentially reasonable second-stage strategy is immediately followed by the declaration that "[i]t is imperative to determine trial counsel's reasons, or lack thereof," for foregoing the use of Dr. Pope's testimony. Id. at 1240. Both of these statements, and indeed the entire discussion of counsel's reasons for his decision, focus on the subjective state of mind of Sallahdin's trial counsel. Since the district court did not find that Barrett had an informed strategic reason for not calling Dr. Pope, neither it nor we reach the issue of whether any such subjective strategy might have been completely unreasonable.

If, as the majority now holds, the objective reasonableness of Barrett's actions is again before us on this appeal, I would again conclude that Sallahdin has overcome Strickland's presumption of objective reasonableness.

As we noted in our prior opinion, there were strong reasons to present Dr. Pope's testimony in the sentencing phase of Sallahdin's trial. In contrast to the vague character evidence offered by Sallahdin's friends and relatives, Dr. Pope's testimony "regarding the potential of steroid use to cause severe personality changes in the user could have explained how Sallahdin could have been transformed from an allegedly mild-mannered, law-abiding individual into a person capable of committing the brutal murder with which he was found guilty." Sallahdin I, 275 F.3d at 1239. The record on remand indicates that Dr. Pope's testimony would also have corroborated Sallahdin's claims both regarding the timing of his last steroid use and regarding his severe withdrawal effects.

All of the reasons that a competent lawyer might have relied upon in deciding not to call Dr. Pope to the stand are either inapplicable to the facts of this case or are objectively unreasonable in the context of those same facts. Lack of funding to cover Dr. Pope's expenses or Dr. Pope's unavailability on short notice might have provided a reasonable justification for not calling him to the stand, but the district court ruled out both these possibilities, and those factual findings are not clearly erroneous.

The majority's first preferred justification — that the jury might have viewed Sallahdin's steroid use in a negative light — does not qualify as a reasonable justification for Barrett's actions. The fact that the jury returned a guilty verdict after a mere 52 minutes of deliberation indicates to a reasonable lawyer that the jurors already had a significantly negative impression of Sallahdin. Adding the fact that the defendant was a long-term user of illegal steroids could hardly, by itself, have lowered him in their eyes. That Sallahdin's steroid use was illegal and that his past experience with the side-effects might have made his continued steroid use reckless may weaken, but does not eliminate, the potential redemptive effect that steroid evidence could have had. The central premise of the majority's proposal — that a reasonable lawyer might have thought Sallahdin's chances of avoiding the death penalty might have been better when he appeared merely to be an intrinsically dangerous random murderer than if he were presented to the jury as a drug-user whose illegal (and now terminated) steroid habit provoked an unanticipated and episodic trigger for his sudden violence — contorts the concept of "reasonableness" beyond recognition.

Neither of the majority's second or third alternatives — that Sallahdin's credibility problems might have undercut his steroid withdrawal claims, or that the prosecution's rebuttal testimony regarding the timing of Sallahdin's last steroid dose would have called into question the force of his withdrawal claims — are objectively reasonable bases for excluding Dr. Pope. The impact of Dr. Pope's scientific conclusions might well have been weakened by the jurors' skepticism toward Sallahdin,2 but even a weakened mitigation argument would have been better than none at all. What is more, the record below indicates that Dr. Pope would have actually corroborated Sallahdin's claims both as to when he last took steroids and as to his symptoms during withdrawal. In any event, Sallahdin's steroid use could have been put into evidence by other witnesses, and the importance of that use could have been explained by Dr. Pope, avoiding entirely the need to call Sallahdin at the penalty phase or to require him directly to contradict his own testimony from the guilt phase of the trial.

Our opinion in Sallahdin I mentioned the possibility that Barrett's actions might have been driven by a residual doubt strategy.3 The circumstances of the trial show, however, that such an approach would not have been objectively reasonable. Although Sallahdin had taken the stand during the guilt phase to claim he was innocent, the jury returned its guilty verdict after deliberating for less than an hour, and all indications in the record are that the prosecution's evidence of guilt was overwhelming. Residual doubt, under such circumstances, was not a plausible tactic to justify Barrett's exclusion of what might have been Sallahdin's strongest mitigation evidence. Further, a steroid defense that could have been presented without the need to call Sallahdin at the penalty phase would not have necessarily been inconsistent with a reasonable doubt strategy as it could have been presented as an alternative basis for avoiding the death penalty. That is, if the jury retained reasonable doubt it should give Sallahdin life imprisonment and even if they did not harbor residual doubt they should still consider life imprisonment because steroid withdrawal would have caused the murder to be atypical and not likely of repetition.

In sum, I see no objectively reasonable justification for excluding Dr. Pope's testimony from the sentencing phase of Sallahdin's trial. We have previously concluded that the exclusion of this testimony was prejudicial. Therefore, I conclude that Barrett's performance as defense counsel was constitutionally deficient, and that his habeas petitions should be granted as to the punishment stage of the trial.

I would affirm the district court's conclusion that there was no evidence of an informed strategic strategy not to call Dr. Pope, and I would reaffirm, if necessary, the prior judgment of this court in Sallahdin I that the decision not to call Dr. Pope was not objectively reasonable. I respectfully dissent and would affirm the district court's grant of habeas corpus here requiring that Sallahdin be given a new penalty phase trial.