Justice BREYER delivered the opinion of the Court.

In Martinez v. Ryan, 566 U.S. 1, 132 S.Ct. 1309, 182 L.Ed.2d 272 (2012), we considered the right of a state prisoner to raise, in a federal habeas corpus proceeding, a claim of ineffective assistance of trial counsel. In that case an Arizona procedural rule required a defendant convicted at trial to raise a claim of ineffective assistance of trial counsel during his first state collateral review proceeding — or lose the claim. The defendant in Martinez did not comply with the state procedural rule. But he argued that the federal habeas court should excuse his state procedural failing, on the ground that he had good "cause" for not raising the claim at the right time, namely that, not only had he lacked effective counsel during trial, but also he lacked effective counsel during his first state collateral review proceeding.

We held that lack of counsel on collateral review might excuse defendant's state law procedural default. We wrote:

At the same time we qualified our holding. We said that the holding applied where state procedural law said that "claims of ineffective assistance of trial counsel must be raised in an initial-review collateral proceeding." Ibid. (emphasis added).

In this case Texas state law does not say "must." It does not on its face require a defendant initially to raise an ineffective-assistance-of-trial-counsel claim in a state collateral review proceeding. Rather, that law appears at first glance to permit (but not require) the defendant initially to raise a claim of ineffective assistance of trial counsel on direct appeal. The structure and design of the Texas system in actual operation, however, make it "virtually impossible" for an ineffective assistance claim to be presented on direct review. See Robinson v. State, 16 S.W.3d 808, 810-811 (Tex.Crim.App.2000). We must now decide whether the Martinez exception applies in this procedural regime. We conclude that it does.

I

A Texas state court jury convicted petitioner, Carlos Trevino, of capital murder. After a subsequent penalty-phase hearing, the jury found that Trevino "would commit criminal acts of violence in the future which would constitute a continuing threat to society," that he "actually caused the death of Linda Salinas or, if he did not actually cause her death, he intended to kill her or another, or he anticipated a human life would be taken," and that "there were insufficient mitigating circumstances to warrant a sentence of life imprisonment" rather than death. 449 Fed. Appx. 415, 418 (C.A.5 2011). The judge consequently imposed a sentence of death.

Eight days later the judge appointed new counsel to handle Trevino's direct appeal. App. 1, 3. Seven months after sentencing, when the trial transcript first became available, that counsel filed an appeal. The Texas Court of Criminal Appeals then considered and rejected Trevino's appellate claims. Trevino's appellate counsel did not claim that Trevino's trial counsel had been constitutionally ineffective during the penalty phase of the trial court proceedings. Id., at 12-24.

About six months after sentencing, the trial judge appointed Trevino a different new counsel to seek state collateral relief. As Texas' procedural rules provide, that third counsel initiated collateral proceedings while Trevino's appeal still was in progress. This new counsel first sought postconviction relief (through collateral review) in the trial court itself. After a hearing, the trial court denied relief; and the Texas Court of Criminal Appeals affirmed that denial. Id., at 25-26, 321-349. Trevino's postconviction claims included a claim that his trial counsel was constitutionally ineffective during the penalty phase of Trevino's trial, but it did not include a claim that trial counsel's ineffectiveness consisted in part of a failure adequately to investigate and to present mitigating circumstances during the penalty phase of Trevino's trial. Id., at 321-349; see Wiggins v. Smith, 539 U.S. 510, 523, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (counsel's failure to investigate and present mitigating circumstances deprived defendant of effective assistance of counsel).

Trevino then filed a petition in federal court seeking a writ of habeas corpus. The Federal District Court appointed another new counsel to represent him. And that counsel claimed for the first time that Trevino had not received constitutionally effective counsel during the penalty phase of his trial in part because of trial counsel's failure to adequately investigate and present mitigating circumstances during the penalty phase. App. 438, 456-478. Federal habeas counsel pointed out that Trevino's trial counsel had presented only one witness at the sentencing phase, namely Trevino's aunt. The aunt had testified that Trevino had had a difficult upbringing, that his mother had an alcohol problem, that his family was on welfare, and that he had dropped out of high school. She had added that Trevino had a child, that he was good with children, and that he was not violent. Id., at 285-291.

Federal habeas counsel then told the federal court that Trevino's trial counsel should have found and presented at the penalty phase other mitigating matters that his own investigation had brought to light. These included, among other things, that Trevino's mother abused alcohol while she was pregnant with Trevino, that Trevino weighed only four pounds at birth, that throughout his life Trevino suffered the deleterious effects of Fetal Alcohol Syndrome, that as a child Trevino had suffered numerous head injuries without receiving adequate medical attention, that Trevino's mother had abused him physically and emotionally, that from an early age Trevino was exposed to, and abused, alcohol and drugs, that Trevino had attended school irregularly and performed poorly, and that Trevino's cognitive abilities were impaired. Id., at 66-67.

The federal court stayed proceedings to permit Trevino to raise this claim in state court. The state court held that because Trevino had not raised this claim during his initial postconviction proceedings, he had procedurally defaulted the claim, id., at 27-28; and the Federal District Court then denied Trevino's ineffective-assistance-of-trial-counsel claim, id., at 78-79. The District Court concluded in relevant part that, despite the fact that "even the most minimal investigation ... would have revealed a wealth of additional mitigating evidence," an independent and adequate state ground (namely Trevino's failure to raise the issue during his state postconviction proceeding) barred the federal habeas court from considering the ineffective-assistance-of-trial-counsel claim. Id., at 131-132. See Coleman v. Thompson, 501 U.S. 722, 729-730, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991).

Trevino appealed. The Fifth Circuit, without considering the merits of Trevino's ineffective-assistance-of-trial-counsel claim, agreed with the District Court that an independent, adequate state ground, namely Trevino's procedural default, barred its consideration. 449 Fed.Appx., at 426. Although the Circuit decided Trevino's case before this Court decided Martinez, the Fifth Circuit's reasoning in a later case, Ibarra v. Thaler, 687 F.3d 222 (2012), makes clear that the Fifth Circuit would have found that Martinez would have made no difference.

That is because in Ibarra the Circuit recognized that Martinez had said that its good-cause exception applies where state law says that a criminal defendant must initially raise his claim of ineffective assistance of trial counsel in initial state collateral review proceedings. 687 F.3d, at 225-226. Texas law, the Circuit pointed out, does not say explicitly that the defendant must initially raise the claim in state collateral review proceedings. Rather Texas law on its face appears to permit a criminal defendant to raise such a claim on direct appeal. Id., at 227. And the Circuit held that that fact means that Martinez does not apply in Texas. 687 F.3d, at 227. Since the Circuit's holding in Ibarra (that Martinez does not apply in Texas) would similarly govern this case, we granted certiorari here to determine whether Martinez applies in Texas.

II

A

We begin with Martinez. We there recognized the historic importance of federal habeas corpus proceedings as a method for preventing individuals from being held in custody in violation of federal law. Martinez, 566 U.S., at ___, 132 S.Ct., at 1315-1316. See generally Preiser v. Rodriguez, 411 U.S. 475, 484-485, 93 S.Ct. 1827, 36 L.Ed.2d 439 (1973). In general, if a convicted state criminal defendant can show a federal habeas court that his conviction rests upon a violation of the Federal Constitution, he may well obtain a writ of habeas corpus that requires a new trial, a new sentence, or release.

We similarly recognized the importance of federal habeas corpus principles designed to prevent federal courts from interfering with a State's application of its own firmly established, consistently followed, constitutionally proper procedural rules. Martinez, supra, at ___, 132 S.Ct., at 1315-1316. Those principles have long made clear that a conviction that rests upon a defendant's state law "procedural default" (for example, the defendant's failure to raise a claim of error at the time or in the place that state law requires), normally rests upon "an independent and adequate state ground." Coleman, 501 U.S., at 729-730, 111 S.Ct. 2546. And where a conviction rests upon such a ground, a federal habeas court normally cannot consider the defendant's federal constitutional claim. Ibid.; see Martinez, 566 U.S., at ___, 132 S.Ct., at 1315-1316.

At the same time, we pointed out that "[t]he doctrine barring procedurally defaulted claims from being heard is not without exceptions. A prisoner may obtain federal review of a defaulted claim by showing cause for the default and prejudice from a violation of federal law." Id., at ___, 132 S.Ct., at 1316. And we turned to the issue directly before the Court: whether Martinez had shown "cause" to excuse his state procedural failing. Id., at ___, 132 S.Ct., at 1320-1321.

Martinez argued that his lawyer should have raised, but did not raise, his claim of ineffective assistance of trial counsel during state collateral review proceedings. Id., at ___, 132 S.Ct., at 1214-1315. He added that this failure, itself amounting to ineffective assistance, was the "cause" of, and ought to excuse, his procedural default. Id., at ___, 132 S.Ct., at 1314-1315. But this Court had previously held that "[n]egligence on the part of a prisoner's postconviction attorney does not qualify as `cause,'" primarily because a "principal" such as the prisoner, "bears the risk of negligent conduct on the part of his agent," the attorney. Maples v. Thomas, 565 U.S. ___, ___, 132 S.Ct. 912, 922, 181 L.Ed.2d 807 (2012) (quoting Coleman, supra, at 753-754, 111 S.Ct. 2546; emphasis added). Martinez, in effect, argued for an exception to Coleman's broad statement of the law.

We ultimately held that a "narrow exception" should "modify the unqualified statement in Coleman that an attorney's ignorance or inadvertence in a postconviction proceeding does not qualify as cause to excuse a procedural default." Martinez, 566 U.S., at ___, 132 S.Ct., at 1315. We did so for three reasons. First, the "right to the effective assistance of counsel at trial is a bedrock principle in our justice system.... Indeed, the right to counsel is the foundation for our adversary system." Id., at ___, 132 S.Ct., at 1317.

Second, ineffective assistance of counsel on direct appellate review could amount to "cause," excusing a defendant's failure to raise (and thus procedurally defaulting) a constitutional claim. Id., at ___, 132 S.Ct, at 1316-1317. But States often have good reasons for initially reviewing claims of ineffective assistance of trial counsel during state collateral proceedings rather than on direct appellate review. Id., at ___, 132 S.Ct., at 1317-1318. That is because review of such a claim normally requires a different attorney, because it often "depend[s] on evidence outside the trial record," and because efforts to expand the record on direct appeal may run afoul of "[a]bbreviated deadlines," depriving the new attorney of "adequate time... to investigate the ineffective-assistance claim." Id., at ___, 132 S.Ct., at 1318.

Third, where the State consequently channels initial review of this constitutional claim to collateral proceedings, a lawyer's failure to raise an ineffective-assistance-of-trial-counsel claim during initial-review collateral proceedings, could (were Coleman read broadly) deprive a defendant of any review of that claim at all. Martinez, supra, at ___, 132 S.Ct., at 1316.

We consequently read Coleman as containing an exception, allowing a federal habeas court to find "cause," thereby excusing a defendant's procedural default, where (1) the claim of "ineffective assistance of trial counsel" was a "substantial" claim; (2) the "cause" consisted of there being "no counsel" or only "ineffective" counsel during the state collateral review proceeding; (3) the state collateral review proceeding was the "initial" review proceeding in respect to the "ineffective-assistance-of-trial-counsel claim"; and (4) state law requires that an "ineffective assistance of trial counsel [claim] ... be raised in an initial-review collateral proceeding." Martinez, supra, at ___, 132 S.Ct., at 1318-1319, 1320-1321.

B

Here state law differs from that in Martinez in respect to the fourth requirement. Unlike Arizona, Texas does not expressly require the defendant to raise a claim of ineffective assistance of trial counsel in an initial collateral review proceeding. Rather Texas law on its face appears to permit (but not require) the defendant to raise the claim on direct appeal. Does this difference matter?

1

Two characteristics of the relevant Texas procedures lead us to conclude that it should not make a difference in respect to the application of Martinez. First, Texas procedure makes it "virtually impossible for appellate counsel to adequately present an ineffective assistance [of trial counsel] claim" on direct review. Robinson, 16 S.W.3d, at 810-811. As the Texas Court of Criminal Appeals itself has pointed out, "the inherent nature of most ineffective assistance" of trial counsel "claims" means that the trial court record will often fail to "contai[n] the information necessary to substantiate" the claim. Ex parte Torres, 943 S.W.2d 469, 475 (1997) (en banc).

As the Court of Criminal Appeals has also noted, a convicted defendant may make a motion in the trial court for a new trial in order to develop the record on appeal. See Reyes v. State, 849 S.W.2d 812, 816 (1993). And, in principle, the trial court could, in connection with that motion, allow the defendant some additional time to develop a further record. Ibid. But that motion-for-new-trial "vehicle is often inadequate because of time constraints and because the trial record has generally not been transcribed at this point." Torres, supra, at 475. See Tex. Rule App. Proc. 21.4 (2013) (motion for a new trial must be made within 30 days of sentencing); Rules 21.8(a), (c) (trial court must dispose of motion within 75 days of sentencing); Rules 35.2(b), 35.3(c) (transcript must be prepared within 120 days of sentencing where a motion for a new trial is filed and this deadline may be extended). Thus, as the Court of Criminal Appeals has concluded, in Texas "a writ of habeas corpus" issued in state collateral proceedings ordinarily "is essential to gathering the facts necessary to ... evaluate ... [ineffective-assistance-of-trial-counsel] claims." Torres, supra, at 475. See Robinson, supra, at 810-811 (noting that there is "not generally a realistic opportunity to adequately develop the record for appeal in post-trial motions" and that "[t]he time requirements for filing and presenting a motion for new trial would have made it virtually impossible for appellate counsel to adequately present an ineffective assistance claim to the trial court").

See also Thompson v. State, 9 S.W.3d 808, 813-814, and n. 6 (Tex.Crim.App. 1999) ("[I]n the vast majority of cases, the undeveloped record on direct appeal will be insufficient for an appellant to satisfy the dual prongs of Strickland [v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984)]"; only "[r]arely will a reviewing court be provided the opportunity to make its determination on direct appeal with a record capable of providing a fair evaluation of the merits of the claim..."); Goodspeed v. State, 187 S.W.3d 390, 392 (Tex.Crim.App.2005) (similar); Andrews v. State, 159 S.W.3d 98, 102-103 (Tex.Crim.App.2005) (similar); Ex parte Brown, 158 S.W.3d 449, 453 (Tex.Crim. App.2005) (per curiam) (similar); Jackson v. State, 973 S.W.2d 954, 957 (Tex.Crim. App.1998) (per curiam) (similar). See also 42 G. Dix & J. Schmolesky, Texas Practice Series § 29:76, pp. 844-845 (3d ed. 2011) (hereinafter Texas Practice) (explaining that "[o]ften" the requirement that a claim of ineffective assistance of trial counsel be supported by a record containing direct evidence of why counsel acted as he did "will require that the claim... be raised in postconviction habeas proceedings where a full record on the matter can be raised").

This opinion considers whether, as a systematic matter, Texas affords meaningful review of a claim of ineffective assistance of trial counsel. The present capital case illustrates why it does not. The trial court appointed new counsel for Trevino eight days after sentencing. Counsel thus had 22 days to decide whether, and on what grounds, to make a motion for a new trial. She then may have had an additional 45 days to provide support for the motion but without the help of a transcript (which did not become available until much later — seven months after the trial). It would have been difficult, perhaps impossible, within that time frame to investigate Trevino's background, determine whether trial counsel had adequately done so, and then develop evidence about additional mitigating background circumstances. See Reyes, supra, at 816 ("[M]otions for new trial [must] be supported by affidavit ... specifically showing the truth of the grounds of attack").

Second, were Martinez not to apply, the Texas procedural system would create significant unfairness. That is because Texas courts in effect have directed defendants to raise claims of ineffective assistance of trial counsel on collateral, rather than on direct, review. As noted, they have explained why direct review proceedings are likely inadequate. See supra, at 1918-1919. They have held that failure to raise the claim on direct review does not bar the defendant from raising the claim in collateral proceedings. See, e.g., Robinson, 16 S.W.3d, at 813; Ex parte Duffy, 607 S.W.2d 507, 512-513 (Tex.Crim.App.1980) (overruled on other grounds by Hernandez v. State, 988 S.W.2d 770 (Tex.Crim.App. 1999)). They have held that the defendant's decision to raise the claim on direct review does not bar the defendant from also raising the claim in collateral proceedings. See, e.g., Lopez v. State, 343 S.W.3d 137, 143 (Tex.Crim.App.2011); Torres, supra, at 475. They have suggested that appellate counsel's failure to raise the claim on direct review does not constitute "ineffective assistance of counsel." See Sprouse v. State, No. AP-74933, 2007 WL 283152, *7 (Tex.Crim.App., Jan. 31, 2007) (unpublished). And Texas' highest criminal court has explicitly stated that "[a]s a general rule" the defendant "should not raise an issue of ineffective assistance of counsel on direct appeal," but rather in collateral review proceedings. Mata v. State, 226 S.W.3d 425, 430, n. 14 (2007) (internal quotation marks omitted). See Robinson, supra, at 810 ("[A] post-conviction writ proceeding, rather than a motion for new trial, is the preferred method for gathering the facts necessary to substantiate" an ineffective-assistance-of-trial-counsel claim).

The criminal bar, not surprisingly, has taken this strong judicial advice seriously. See Guidelines and Standards for Texas Capital Counsel, 69 Tex. B.J. 966, 977, Guideline 12.2(B)(1)(d) (2006) ("[S]tate habeas corpus is the first opportunity for a capital client to raise challenges to the effectiveness of trial or direct appeal counsel"). Texas now can point to only a comparatively small number of cases in which a defendant has used the motion-for-a-new-trial mechanism to expand the record on appeal and then received a hearing on his ineffective-assistance-of-trial-counsel claim on direct appeal. Brief for Respondent 35-36, and n. 6 (citing, inter alia, State v. Morales, 253 S.W.3d 686, 689-691 (Tex.Crim.App.2008); Robertson v. State, 187 S.W.3d 475, 480-481 (Tex.Crim.App. 2006)). And, of those, precisely one case involves trial counsel's investigative failures of the kind at issue here. See Armstrong v. State, No. AP-75706, 2010 WL 359020 (Tex.Crim.App., Jan. 27, 2010) (unpublished). How could federal law deny defendants the benefit of Martinez solely because of the existence of a theoretically available procedural alternative, namely direct appellate review, that Texas procedures render so difficult, and in the typical case all but impossible, to use successfully, and which Texas courts so strongly discourage defendants from using?

Respondent argues that Texas courts enforce the relevant time limits more flexibly than we have suggested. Sometimes, for example, an appellate court can abate an appeal and remand the case for further record development in the trial court. See Cooks v. State, 240 S.W.3d 906 (Tex.Crim. App.2007). But the procedural possibilities to which Texas now points seem special, limited in their application, and, as far as we can tell, rarely used. See 43A Texas Practice § 50:15, at 636-639; 43B id., § 56:235, at 607-609. Cooks, for example, the case upon which respondent principally relies, involved a remand for further record development, but in circumstances where the lower court wrongly failed to give a defendant new counsel in time to make an ordinary new trial motion. 240 S.W.3d, at 911. We do not believe that this, or other, special, rarely used procedural possibilities can overcome the Texas courts' own well-supported determination that collateral review normally constitutes the preferred — and indeed as a practical matter, the only — method for raising an ineffective-assistance-of-trial-counsel claim.

Respondent further argues that there is no equitable problem to be solved in Texas because if counsel fails to bring a substantial claim of ineffective assistance of trial counsel on direct appeal, the ineffectiveness of appellate counsel may constitute cause to excuse the procedural default. See Murray v. Carrier, 477 U.S. 478, 106 S.Ct. 2639, 91 L.Ed.2d 397 (1986). But respondent points to no case in which such a failure by appellate counsel has been deemed constitutionally ineffective. And that lack of authority is not surprising given the fact that the Texas Court of Criminal Appeals has directed defendants to bring such claims on collateral review.

2

For the reasons just stated, we believe that the Texas procedural system — as a matter of its structure, design, and operation — does not offer most defendants a meaningful opportunity to present a claim of ineffective assistance of trial counsel on direct appeal. What the Arizona law prohibited by explicit terms, Texas law precludes as a matter of course. And, that being so, we can find no significant difference between this case and Martinez. The very factors that led this Court to create a narrow exception to Coleman in Martinez similarly argue for the application of that exception here.

The right involved — adequate assistance of counsel at trial — is similarly and critically important. In both instances practical considerations, such as the need for a new lawyer, the need to expand the trial court record, and the need for sufficient time to develop the claim, argue strongly for initial consideration of the claim during collateral, rather than on direct, review. See Martinez, 566 U.S., at ___, 132 S.Ct., at 1318; see also Massaro v. United States, 538 U.S. 500, 505, 123 S.Ct. 1690, 155 L.Ed.2d 714 (2003). In both instances failure to consider a lawyer's "ineffectiveness" during an initial-review collateral proceeding as a potential "cause" for excusing a procedural default will deprive the defendant of any opportunity at all for review of an ineffective-assistance-of-trial-counsel claim. See Martinez, supra, at ___, 132 S.Ct., at 1316.

Thus, for present purposes, a distinction between (1) a State that denies permission to raise the claim on direct appeal and (2) a State that in theory grants permission but, as a matter of procedural design and systemic operation, denies a meaningful opportunity to do so is a distinction without a difference. In saying this, we do not (any more than we did in Martinez) seek to encourage States to tailor direct appeals so that they provide a fuller opportunity to raise ineffective-assistance-of-trial-counsel claims. That is a matter for the States to decide. And, as we have said, there are often good reasons for hearing the claim initially during collateral proceedings.

III

For these reasons, we conclude that where, as here, state procedural framework, by reason of its design and operation, makes it highly unlikely in a typical case that a defendant will have a meaningful opportunity to raise a claim of ineffective assistance of trial counsel on direct appeal, our holding in Martinez applies:

Given this holding, Texas submits that its courts should be permitted, in the first instance, to decide the merits of Trevino's ineffective-assistance-of-trial-counsel claim. Brief for Respondent 58-60. We leave that matter to be determined on remand. Likewise, we do not decide here whether Trevino's claim of ineffective assistance of trial counsel is substantial or whether Trevino's initial state habeas attorney was ineffective.

For these reasons we vacate the Fifth Circuit's judgment and remand the case for further proceedings consistent with this opinion.

It is so ordered.

Chief Justice ROBERTS, with whom Justice ALITO joins, dissenting.

In our federal system, the "state courts are the principal forum for asserting constitutional challenges to state convictions." Harrington v. Richter, 562 U.S. ___, ___, 131 S.Ct. 770, 787, 178 L.Ed.2d 624 (2011). "Federal courts sitting in habeas," we have said, "are not an alternative forum for trying ... issues which a prisoner made insufficient effort to pursue in state proceedings." Williams v. Taylor, 529 U.S. 420, 437, 120 S.Ct. 1479, 146 L.Ed.2d 435 (2000). This basic principle reflects the fact that federal habeas review "`intrudes on state sovereignty to a degree matched by few exercises of federal judicial authority.'" Richter, supra, at ___, 131 S.Ct., at 787 (quoting Harris v. Reed, 489 U.S. 255, 282, 109 S.Ct. 1038, 103 L.Ed.2d 308 (1989) (KENNEDY, J., dissenting)).

In order to prevent circumvention of the state courts and the unjustified intrusion on state sovereignty that results, we have held that "a state prisoner [who] fails to exhaust state remedies ... [or] has failed to meet the State's procedural requirements for presenting his federal claims" will not be entitled to federal habeas relief unless he can show "cause" to excuse his default. Coleman v. Thompson, 501 U.S. 722, 732, 750, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991). There is an exception to that rule where "failure to consider the claims will result in a fundamental miscarriage of justice," ibid.; that exception is not at issue here.

Cause comes in different forms, but the one relevant here is attorney error. We recognized in Coleman that "[w]here a [habeas] petitioner defaults a claim as a result of the denial of the right to effective assistance of counsel, the State, which is responsible for the denial as a constitutional matter, must bear the cost of any resulting default." Id., at 754, 111 S.Ct. 2546. But we simultaneously recognized that "[a] different allocation of costs is appropriate in those circumstances where the State has no responsibility to ensure that the petitioner was represented by competent counsel." Ibid. In that situation, we held, "it is the petitioner who must bear the burden of a failure to follow state procedural rules." Ibid. Because the error in Coleman occurred during state postconviction proceedings, a point at which the habeas petitioner had no constitutional right to counsel, the petitioner had to bear the cost of his default. Id., at 757, 111 S.Ct. 2546.

Last Term, in Martinez v. Ryan, we announced a "narrow exception" to Coleman's "unqualified statement ... that an attorney's ignorance or inadvertence in a postconviction proceeding does not qualify as cause to excuse a procedural default." 566 U.S. 1, ___, 132 S.Ct. 1309, 1315, 182 L.Ed.2d 272 (2012). In Martinez, Arizona law did not allow defendants to raise ineffective assistance of counsel claims on direct appeal; they could only raise such claims in state collateral proceedings. Id., at ___, 132 S.Ct., at 1313-1314. We held that while Arizona was free to structure its state court procedures in this way, its "decision is not without consequences for the State's ability to assert a procedural default in later proceedings." Id., at ___, 132 S.Ct., at 1318. "By deliberately choosing to move trial-ineffectiveness claims outside of the direct-appeal process, where counsel is constitutionally guaranteed, the State significantly diminishes prisoners' ability to file such claims." Ibid. Thus, "within the context of this state procedural framework," attorney error would qualify as cause to excuse procedural default if it occurred in the first proceeding at which the prisoner was "allow[ed]" to raise his trial ineffectiveness claim. Id., at ___, ___, 132 S.Ct., at 1318, 1319-1320.

We were unusually explicit about the narrowness of our decision: "The holding in this case does not concern attorney errors in other kinds of proceedings," and "does not extend to attorney errors in any proceeding beyond the first occasion the State allows a prisoner to raise a claim of ineffective assistance at trial." Id., at ___-___, 132 S.Ct., at 1320. "Our holding here addresses only the constitutional claims presented in this case, where the State barred the defendant from raising the claims on direct appeal." Id., at ___, 132 S.Ct., at 1320. In "all but the limited circumstances recognized here," we said, "[t]he rule of Coleman governs." Id., at ___, 132 S.Ct., at 1320.

This aggressively limiting language was not simply a customary nod to the truism that "we decide only the case before us." Upjohn Co. v. United States, 449 U.S. 383, 396, 101 S.Ct. 677, 66 L.Ed.2d 584 (1981). It was instead an important part of our explanation for why "[t]his limited qualification to Coleman does not implicate the usual concerns with upsetting reliance interests protected by stare decisis principles." Martinez, supra, at ___, 132 S.Ct., at 1319. The fact that the exception was clearly delineated ensured that the Coleman rule would remain administrable. And because States could readily anticipate how such a sharply defined exception would apply to various procedural frameworks, the exception could be reconciled with our concerns for comity and equitable balancing that led to Coleman's baseline rule in the first place. See Coleman, supra, at 750-751, 111 S.Ct. 2546. The States had a clear choice, which they could make with full knowledge of the consequences: If a State "deliberately cho[se] to move trial-ineffectiveness claims outside of the direct-appeal process" through a "decision to bar defendants from raising" them there, then — and only then — would "counsel's ineffectiveness in an initial-review collateral proceeding qualif[y] as cause for a procedural default." Martinez, 566 U.S., at ___, ___, 132 S.Ct., at 1318, 1320.

Today, with hardly a mention of these concerns, the majority throws over the crisp limit we made so explicit just last Term. We announced in Martinez that the exception applies "where the State barred the defendant from raising the claims on direct appeal." Id., at ___, 132 S.Ct., at 1320. But today, the Court takes all the starch out of its rule with an assortment of adjectives, adverbs, and modifying clauses: Martinez's "narrow exception" now applies whenever the "state procedural framework, by reason of its design and operation, makes it highly unlikely in a typical case that a defendant will have a meaningful opportunity" to raise his claim on direct appeal. Ante, at 1921.

The questions raised by this equitable equation are as endless as will be the state-by-state litigation it takes to work them out. We are not told, for example, how meaningful is meaningful enough, how meaningfulness is to be measured, how unlikely highly unlikely is, how often a procedural framework's "operation" must be reassessed, or what case qualifies as the "typical" case. Take just this last example: The case before us involved a jury trial (hardly typical), a capital conviction (even less typical), and — as the majority emphasizes — a particular species of ineffectiveness claim that depends on time-consuming investigation of personal background and other mitigating circumstances. Ante, at 1919. Yet the majority holds it up, apparently, as a case that is typical in the relevant sense, saying that "[t]he present capital case illustrates" the "systematic" working of Texas's procedural framework. Ibid.

Given that the standard is so opaque and malleable, the majority cannot describe the exception applied here as narrow, and does not do so. Gone are the repeated words of limitation that characterized the Martinez opinion. Gone too is the clear choice that Martinez gave the States about how to structure their criminal justice systems. Now, the majority offers them a gamble: If a State allows defendants to bring ineffectiveness claims both on direct appeal and in postconviction proceedings, then a prisoner might have to comply with state procedural requirements in order to preserve the availability of federal habeas review, if a federal judge decides that the state system gave the defendant (or enough other "typical" defendants) a sufficiently meaningful opportunity to press his claim.

This invitation to litigation will, in precisely the manner that Coleman foreclosed, "`frustrate both the States' sovereign power to punish offenders and their good-faith attempts to honor constitutional rights.'" Coleman, 501 U.S., at 748, 111 S.Ct. 2546 (quoting Engle v. Isaac, 456 U.S. 107, 128, 102 S.Ct. 1558, 71 L.Ed.2d 783 (1982)). In what I suspect (though cannot know) will be a broad swath of cases, the Court's approach will excuse procedural defaults that, under Coleman, should preclude federal review. But even in cases where federal courts ultimately decide that the habeas petitioner cannot establish cause under the new standard, the years of procedural wrangling it takes to reach that decision will themselves undermine the finality of sentences necessary to effective criminal justice. Because that approach is inconsistent with Coleman, Martinez itself, and the principles of equitable discretion and comity at the heart of both, I respectfully dissent.

Justice SCALIA, with whom Justice THOMAS joins, dissenting.

I dissent for the reasons set forth in my dissent in Martinez v. Ryan, 566 U.S. 1, 132 S.Ct. 1309, 182 L.Ed.2d 272 (2012). That opinion sought to minimize the impact of its novel holding as follows:

I wrote in my dissent:

The Court says today: