delivered the opinion of the Court.
We granted certoriari in this case to consider whether a federal habeas petitioner can show cause for a procedural default by establishing that competent defense counsel inad*482vertently failed to raise the substantive claim of error rather than deliberately withholding it for tactical reasons.
I
Respondent Clifford Carrier was convicted of rape and abduction by a Virginia jury in 1977. Before trial, respondent’s court-appointed counsel moved for discovery of the victim’s statements to police describing “her assailants, the vehicle the assailants were driving, and the location of where the alleged rape took place.” 2 Record 11. The presiding judge denied the motion by letter to counsel after examining the statements in camera and determining that they contained no exculpatory evidence. Id., at 31. Respondent’s counsel made a second motion to discover the victim’s statements immediately prior to trial, which the trial judge denied for the same reason after conducting his own in camera examination. Tr. 151-152.
After respondent was convicted, his counsel filed a notice of appeal to the Virginia Supreme Court assigning seven errors, of which the fifth was:
“Did the trial judge err by not permitting defendant’s counsel to examine the written statements of the victim prior to trial, and during the course of the trial?” 2 Record 83.
Without consulting respondent, counsel subsequently submitted the required petition for appeal but failed to include this claim, notwithstanding that Virginia Supreme Court Rule 5:21 provides that “[ojnly errors assigned in the petition for appeal will be noticed by this Court and no error not so assigned will be admitted as a ground for reversal of a decision below.” The Virginia Supreme Court refused the appeal and this Court denied certiorari. Carrier v. Virginia, 439 U. S. 1076 (1979).
A year later respondent, by this time proceeding pro se, filed a state habeas corpus petition claiming that he had been denied due process of law by the prosecution’s withholding of the victim’s statements. The State sought dismissal of his *483petition on the ground that respondent was barred from presenting his due process discovery claim on collateral review because he failed to raise that claim on appeal. The state ha-beas court dismissed the petition “for the reasons stated in the Motion to Dismiss,” 1 Record, Doc. No. 12, and the Virginia Supreme Court denied certiorari.
Respondent next filed a pro se habeas petition in the District Court for the Eastern District of Virginia, renewing his due process discovery claim as grounds for relief. The State filed a motion to dismiss asserting that respondent’s failure to raise the issue on direct appeal was a procedural default barring federal habeas review under Wainwright v. Sykes, 433 U. S. 72 (1977), and that respondent had not exhausted his state remedies because he could bring an ineffective assistance of counsel claim in the state courts to establish that his procedural default should be excused. 1 Record, Doc. No. 3. The United States Magistrate to whom the case was referred recommended dismissal by virtue of the procedural default and also ruled that respondent had not exhausted his state remedies. In reply to the Magistrate’s report, respondent alleged that his procedural default was “due to ineffective assistance of counsel during the filing of his appeal.” App. 11. The District Court approved the Magistrate’s report, holding the discovery claim barred by the procedural default and indicating that respondent should establish cause for that default in the state courts.
At oral argument on appeal to the Court of Appeals for the Fourth Circuit, respondent abandoned any claim of ineffective assistance of counsel but asserted that counsel had mistakenly omitted his discovery claim from the petition for appeal and that this error was cause for his default. A divided panel of the Court of Appeals reversed and remanded. Carrier v. Hutto, 724 F. 2d 396 (1983). The court construed respondent’s objection to the denial of discovery as having rested throughout on a contention that Brady v. Maryland, 373 U. S. 83 (1963), requires the prosecution to disclose any evidence that might be material to guilt whether or not it is *484exculpatory, and concluded that when respondent’s counsel omitted this discovery claim from the petition for review “the issue was lost for purposes of direct and collateral review.” 724 F. 2d, at 399. The court framed the issue before it as whether “a single act or omission by counsel, insufficient by itself to contravene the sixth amendment, [can] satisfy the ‘cause’ prong of the exception to preclusive procedural default discussed in Waimvright?” Id., at 400. In answering this question, the court drew a dispositive distinction between procedural defaults resulting from deliberate tactical decisions and those resulting from ignorance or inadvertence. Id., at 401. The court determined that only in the latter category does an attorney’s error constitute cause because, whereas a tactical decision implies that counsel has, at worst, “reasonably but incorrectly exercisefd] her judgment,” ignorance or oversight implies that counsel “fail[ed] to exercise it at all, in dereliction of the duty to represent her client.” Ibid. Thus, in order to establish cause a federal habeas petitioner need only satisfy the district court “that the failure to object or to appeal his claim was the product of his attorney’s ignorance or oversight, not a deliberate tactic.” Ibid. Accordingly, the Court of Appeals remanded to the District Court:
“[Although the likelihood of attorney error appears very great in this case, we lack testimony from Carrier’s counsel which might disclose a strategic reason for failing to appeal the Brady issue. The question of counsel’s motivation is one of fact for the district court to resolve upon taking further evidence.” Id., at 402.
The court also ruled that the District Court erred in suggesting that respondent should establish cause for the default in the state courts. “The exhaustion requirement of 28 U. S. C. § 2254 pertains to independent claims for habeas relief, not to the proffer of Waimvright cause and prejudice.” Ibid. Since respondent did not allege ineffective assistance *485of counsel as an independent basis for habeas relief, the case presented no exhaustion question.
The dissenting judge believed that the petition should have been dismissed for failure to exhaust state remedies because respondent had never presented his discovery claim as a denial of due process in the state courts, id., at 403-404 (Hall, J., dissenting), and differed with the majority’s interpretation of the cause standard because “[it] will ultimately allow the exception to swallow the rule.” Id., at 405. The State sought rehearing, and the en banc Court of Appeals adopted the panel majority’s decision, with four judges dissenting. Carrier v. Hutto, 754 F. 2d 520 (1985). We now reverse and remand.
I — I u_i
Wainwright v. Sykes held that a federal habeas petitioner who has failed to comply with a State’s contemporaneous-objection rule at trial must show cause for the procedural default and prejudice attributable thereto in order to obtain review of his defaulted constitutional claim. 433 U. S., at 87. See also Francis v. Henderson, 425 U. S. 536 (1976). In so holding, the Court explicitly rejected the standard described in Fay v. Noia, 372 U. S. 391 (1963), under which a federal habeas court could refuse to review a defaulted claim only if “an applicant ha[d] deliberately by-passed the orderly procedure of the state courts,” id., at 438, by personal waiver of the claim amounting to “ ‘an intentional relinquishment or abandonment of a known right or privilege.’” Id., at 439 (quoting Johnson v. Zerbst, 304 U. S. 458, 464 (1938)). See Wainwright v. Sykes, 433 U. S., at 87-88. At a minimum, then, Wainwright v. Sykes plainly implied that default of a constitutional claim by counsel pursuant to a trial strategy or tactical decision would, absent extraordinary circumstances, bind the habeas petitioner even if he had not personally waived that claim. See id., at 91, n. 14; Reed v. Ross, 468 U. S. 1, 13 (1984). Beyond that, the Court left open “for *486resolution in future decisions the precise definition of the ‘cause’-and-‘prejudice’ standard.” 433 U. S., at 87.
We revisited the cause and prejudice test in Engle v. Isaac, 456 U. S. 107 (1982). Like Waimvright v. Sykes, Engle involved claims that were procedurally defaulted at trial. In seeking to establish cause for their defaults, the prisoners argued that “they could not have known at the time of their trials” of the substantive basis for their constitutional claims, which were premised on In re Winship, 397 U. S. 358 (1970). Engle, 456 U. S., at 130. Without deciding “whether the novelty of a constitutional claim ever establishes cause for a failure to object,” id., at 131, we rejected this contention because we could not conclude that the legal basis for framing the prisoners’ constitutional claims was unavailable at the time. Id., at 133. In language that bears directly on the present case, we said:
“We do not suggest that every astute counsel would have relied upon Winship to assert the unconstitutionally of a rule saddling criminal defendants with the burden of proving an affirmative defense. Every trial presents a myriad of possible claims. Counsel might have overlooked or chosen to omit respondents’ due process argument while pursuing other avenues of defense. We have long recognized, however, that the Constitution guarantees criminal defendants only a fair trial and a competent attorney. It does not insure that defense counsel will recognize and raise every conceivable constitutional claim. Where the basis of a constitutional claim is available, and other defense counsel have perceived and litigated that claim, the demands of comity and finality counsel against labeling alleged unawareness of the objection as a cause for a procedural default.” Id., at 133-134 (footnote omitted).
The thrust of this part of our decision in Engle is unmistakable: the mere fact that counsel failed to recognize the factual or legal basis for a claim, or failed to raise the claim despite recognizing it, does not constitute cause for a procedural de*487fault. At least with respect to defaults that occur at trial, the Court of Appeals’ holding that ignorant or inadvertent attorney error is cause for any resulting procedural default is plainly inconsistent with Engle. It is no less inconsistent with the purposes served by the cause and prejudice standard. That standard rests not only on the need to deter intentional defaults but on a judgment that the costs of federal ha-beas review “are particularly high when a trial default has barred a prisoner from obtaining adjudication of his constitutional claim in the state courts.” Engle, 456 U. S., at 128. Those costs, which include a reduction in the finality of litigation and the frustration of “both the States’ sovereign power to punish offenders and their good-faith attempts to honor constitutional rights,” ibid., are heightened in several respects when a trial default occurs: the default deprives the trial court of an opportunity to correct any error without retrial, detracts from the importance of the trial itself, gives state appellate courts no chance to review trial errors, and “exacts an extra charge by undercutting the State’s ability to enforce its procedural rules.” Id., at 129. Clearly, these considerable costs do not disappear when the default stems from counsel’s ignorance or inadvertence rather than from a deliberate decision, for whatever reason, to withhold a claim.
Indeed, the rule applied by the Court of Appeals would significantly increase the costs associated with a procedural default in many cases. In order to determine whether there was cause for a procedural default, federal habeas courts would routinely be required to hold evidentiary hearings to determine what prompted counsel’s failure to raise the claim in question. While the federal habeas courts would no doubt strive to minimize the burdens to all concerned through the use of affidavits or other simplifying procedures, we are not prepared to assume that these costs would be negligible, particularly since, as we observed in Strickland v. Washington, 466 U. S. 668, 690 (1984), “[intensive scrutiny of counsel. . . could dampen the ardor and impair the independence of defense counsel, discourage the acceptance of assigned cases, *488and undermine the trust between attorney and client.” Nor will it always be easy to classify counsel’s behavior in accordance with the deceptively simple categories propounded by the Court of Appeals. Does counsel act out of “ignorance,” for example, by failing to raise a claim for tactical reasons after mistakenly assessing its strength on the basis of an incomplete acquaintance with the relevant precedent? The uncertain dimensions, of any exception for “inadvertence” or “ignorance” furnish an additional reason for rejecting it.
We think, then, that the question of cause for a procedural default does not turn on whether counsel erred or on the kind of error counsel may have made. So long as a defendant is represented by counsel whose performance is not constitutionally ineffective under the standard established in Strickland v. Washington, supra, we discern no inequity in requiring him to bear the risk of attorney error that results in a procedural default. Instead, we think that the existence of cause for a procedural default must ordinarily turn on whether the prisoner can show that some objective factor external to the defense impeded counsel’s efforts to comply with the State’s procedural rule. Without attempting an exhaustive catalog of such objective impediments to compliance with a procedural rule, we note that a showing that the factual or legal basis for a claim was not reasonably available to counsel, see Reed v. Ross, 468 U. S., at 16, or that “some interference by officials,” Brown v. Allen, 344 U. S. 443, 486 (1953), made compliance impracticable, would constitute cause under this standard.
Similarly, if the procedural default is the result of ineffective assistance of counsel, the Sixth Amendment itself requires that responsibility for the default be imputed to the State, which may not “conduc[t] trials at which persons who face incarceration must defend themselves without adequate legal assistance.” Cuyler v. Sullivan, 446 U. S. 335, 344 (1980). Ineffective assistance of counsel, then, is cause for a procedural default. However, we think that the exhaustion *489doctrine, which is “principally designed to protect the state courts’ role in the enforcement of federal law and prevent disruption of state judicial proceedings,” Rose v. Lundy, 455 U. S. 509, 518 (1982), generally requires that a claim of ineffective assistance be presented to the state courts as an independent claim before it may be used to establish cause for a procedural default. The question whether there is cause for a procedural default does not pose any occasion for applying the exhaustion doctrine when the federal habeas court can adjudicate the question of cause — a question of federal law— without deciding an independent and unexhausted constitutional claim on the merits. But if a petitioner could raise his ineffective assistance claim for the first time on federal ha-beas in order to show cause for a procedural default, the federal habeas court would find itself in the anomalous position of adjudicating an unexhausted constitutional claim for which state court review might still be available. The principle of comity that underlies the exhaustion doctrine would be ill served by a rule that allowed a federal district court “to upset a state court conviction without an opportunity to the state courts to correct a constitutional violation,” Darr v. Burford, 339 U. S. 200, 204 (1950), and that holds true whether an ineffective assistance claim is asserted as cause for a procedural default or denominated as an independent ground for habeas relief.
It is clear that respondent failed to show or even allege cause for his procedural default under this standard for cause, which Engle squarely supports. Respondent argues nevertheless that his case is not controlled by Engle because it involves a procedural default on appeal rather than at trial. Respondent does not dispute, however, that the cause and prejudice test applies to procedural defaults on appeal, as we plainly indicated in Reed v. Ross, 468 U. S., at 11. Reed, which involved a claim that was defaulted on appeal, held that a habeas petitioner could establish cause for a procedural default if his claim is “so novel that its legal basis is not rea*490sonably available to counsel,” id., at 16. That holding would have been entirely unnecessary to the disposition of the prisoner’s claim if the cause and prejudice test were inapplicable to procedural defaults on appeal.
The distinction respondent would have us draw must therefore be made, if at all, in terms of the content of the cause requirement as applied to procedural defaults on appeal. Accordingly, respondent asks us to affirm the Court of Appeals’ judgment on the narrow ground that even if counsel’s ignorance or inadvertence does not constitute cause for a procedural default at trial, it does constitute cause for a procedural default on appeal. In support of this distinction, respondent asserts that the concerns that underlie the cause and prejudice test are not present in the case of defaults on appeal. A default on appeal, he maintains, does not detract from the significance of the trial or from the development of a full trial record, or deprive the trial court of an opportunity to correct error without the need for retrial. Moreover, unlike the rapid pace of trial, in which it is a matter of necessity that counsel’s decisions bind the defendant, “the appellate process affords the attorney time for reflection, research, and full consultation with his client.” Brief for Respondent 19. Finally, respondent suggests that there is no likelihood that an attorney will preserve an objection at trial yet choose to withhold it on appeal in order to “sandbag” the prosecution by raising the claim on federal habeas if relief is denied by the state courts.
These arguments are unpersuasive. A State’s procedural rules serve vital purposes at trial, on appeal, and on state collateral attack. The important role of appellate procedural rules is aptly captured by the Court’s description in Reed v. Ross of the purposes served by the procedural rule at issue there, which required the defendant initially to raise his legal claims on appeal rather than on postconviction review:
“It affords the state courts the opportunity to resolve the issue shortly after trial, while evidence is still avail*491able both to assess the defendant’s claim and to retry the defendant effectively if he prevails in his appeal. See Friendly, Is Innocence Irrelevant? Collateral Attack on Criminal Judgments, 38 U. Chi. L. Rev. 142, 147 (1970). This type of rule promotes not only the accuracy and efficiency of judicial decisions, but also the finality of those decisions, by forcing the defendant to litigate all of his claims together, as quickly after trial as the docket will allow, and while the attention of the appellate court is focused on his case.” 468 U. S., at 10-11.
These legitimate state interests, which are manifestly furthered by the comparable procedural rule at issue in this case, warrant our adherence to the conclusion to which they led the Court in Reed v. Ross — that the cause and prejudice test applies to defaults on appeal as to those at trial.
We likewise believe that the standard for cause should not vary depending on the timing of a procedural default or on the strength of an uncertain and difficult assessment of the relative magnitude of the benefits attributable to the state procedural rules that attach at each successive stage of the judicial process. “Each State’s complement of procedural rules . . . channels], to the extent possible, the resolution of various types of questions to the stage of the judicial process at which they can be resolved most fairly and efficiently.” Id., at 10. It is apparent that the frustration of the State’s interests that occurs when an appellate procedural rule is broken is not significantly diminished when counsel’s breach results from ignorance or inadvertence rather than a deliberate decision, tactical or not, to abstain from raising the claim. Failure to raise a claim on appeal reduces the finality of appellate proceedings, deprives the appellate court of .an opportunity to review trial error, and “undercuts] the State’s ability to enforce its procedural rules.” Engle, 456 U. S., at 129. As with procedural defaults at trial, these costs are imposed on the State regardless of the kind of attorney error that led to the procedural default. Nor do we *492agree that the possibility of “sandbagging” vanishes once a trial has ended in conviction, since appellate counsel might well conclude that the best strategy is to select a few promising claims for airing on appeal, while reserving others for federal habeas review should the appeal be unsuccessful. Moreover, we see little reason why counsel’s failure to detect a colorable constitutional claim should be treated differently from a deliberate but equally prejudicial failure by counsel to raise such a claim. The fact that the latter error can be characterized as a misjudgment, while the former is more easily described as an oversight, is much too tenuous a distinction to justify a regime of evidentiary hearings into counsel’s state of mind in failing to raise a claim on appeal.
The real thrust of respondent’s arguments appears to be that on appeal it is inappropriate to hold defendants to the errors of their attorneys. Were we to accept that proposition, defaults on appeal would presumably be governed by a rule equivalent to Fay v. Noia’s “deliberate bypass” standard, under which only personal waiver by the defendant would require enforcement of a procedural default. We express no opinion as to whether counsel’s decision not to take an appeal at all might require treatment under such a standard, see Wainwright v. Sykes, 433 U. S., at 88, n. 12, but, for the reasons already given, we hold that counsel’s failure to raise a particular claim on appeal is to be scrutinized under the cause and prejudice standard when that failure is treated as a procedural default by the state courts. Attorney error short of ineffective assistance of counsel does not constitute cause for a procedural default even when that default occurs on appeal rather than at trial. To the contrary, cause for a procedural default on appeal ordinarily requires a showing of some external impediment preventing counsel from constructing or raising the claim.
I — I ) — 4 1 — 4
Concurring in the judgment, Justice Stevens contends that our decision today erects an unwarranted procedural *493barrier to the correction through federal habeas corpus of violations of fundamental constitutional rights that have resulted in a miscarriage of justice. The cause and prejudice test, in his view, “must be considered within an overall inquiry into justice,” post, at 504, which requires consideration in every case of the character of the constitutional claim. If the federal right asserted is of “fundamental importance,” post, at 499, or if a violation of that right “calls into question the accuracy of the determination of . . . guilt,” ibid., Justice Stevens would then balance “the nature and strength of the constitutional claim” and “the nature and strength of the state procedural rule that has not been observed.” Post, at 506.
At the outset, it should be noted that this balancing is more apparent than real, for the concurrence makes plain that the controlling consideration must be whether the petitioner was denied “‘fundamental fairness in the state-court proceedings.’” Post, at 506, n. 13 (quoting Rose v. Lundy, 455 U. S. 509, 547, n. 17 (1982) (Stevens, J., dissenting)). And, while Justice Stevens argues at some length that an appellate default should be given less weight than a trial default in applying the balancing process he proposes, it is hard to believe that this distinction would make any difference given his simultaneous insistence on “carefully preserving] the exception which enables the federal writ to grant relief in cases of manifest injustice,” post, at 515 — an exception that he clearly would endorse regardless of the timing of the default.
The effect of such a reworking of the cause and prejudice test would essentially be to dispense with the requirement that the petitioner show cause and instead to focus exclusively on whether there has been a “manifest injustice” or a denial of “fundamental fairness.” We are not told whether this inquiry would require the same showing of actual prejudice that is required by the cause and prejudice test as interpreted in Engle and in United States v. Frady, 456 U. S. 152 (1982), but the thrust of the concurrence leaves little doubt that this is so. The showing of prejudice required under *494Wainwright v. Sykes is significantly greater than that necessary under “the more vague inquiry suggested by the words ‘plain error.’” Engle, 456 U. S., at 135; Frady, supra, at 166. See also Henderson v. Kibbe, 431 U. S. 145, 154 (1977). The habeas petitioner must show “not merely that the errors at. . . trial created a possibility of prejudice, but that they worked to his actual and substantial disadvantage, infecting his entire trial with error of constitutional dimensions.” Frady, supra, at 170. Such a showing of pervasive actual prejudice can hardly be thought to constitute anything other than a showing that the prisoner was denied “fundamental fairness” at trial. Since, for Justice Stevens, a “constitutional claim that implicates ‘fundamental fairness’ . . . compels review regardless of possible procedural defaults,” post, at 501, n. 8, it follows that a showing of prejudice would invariably make a showing of cause unnecessary.
As the concurrence acknowledges, Engle expressly rejected this contention that a showing of actual prejudice “should permit relief even in the absence of cause.” 456 U. S., at 134, n. 43. It may be true that the former Rule 12(b)(2) of the Federal Rules of Criminal Procedure, as interpreted in Shotwell Mfg. Co. v. United States, 371 U. S. 341 (1963), and Davis v. United States, 411 U. S. 233 (1973), treated prejudice as a component of the inquiry into whether there was cause for noncompliance with that rule. But, while the cause and prejudice test adopted in Wainwright v. Sykes finds its antecedents in cases interpreting Rule 12(b)(2), the Court in Wainwright v. Sykes declared that it was applying “the rule of Francis v. Henderson . . . barring federal habeas review absent a showing of ‘cause’ and ‘prejudice’ attendant to a state procedural waiver.” 433 U. S., at 87. In Francis, the Court could not have been clearer that both cause and prejudice must be shown, at least in a habeas corpus proceeding challenging a state court conviction. 425 U. S., at 542 (requiring “not only a showing of ‘cause’ for the defendant’s failure to challenge the composition of the grand *495jury before trial, but also a showing of actual prejudice”). We deal here with habeas review of a state court conviction, and at least three decisions of this Court — Francis, Sykes, and Engle — axe unambiguously contrary to the approach taken in the concurrence. We are unprepared, in the face of this weight of authority and in view of the principles of comity and finality these decisions reflect, to reduce the cause requirement to the vestigial role Justice Stevens envisions for it.
Moreover, although neither Francis nor Wainwright v. Sykes involved a constitutional claim that directly called into question the accuracy of the determination of the petitioner’s guilt, the defaulted claims in Engle, no less than respondent’s claim in this case, did involve issues bearing on the reliability of the verdict. In re Winship, 397 U. S. 358 (1970), which was “the basis for [the prisoners’] constitutional claim” in Engle, supra, at 131, holds that “the Due Process Clause protects the accused against conviction except upon proof beyond a reasonable doubt of every fact necessary to constitute the crime with which he is charged.” Winship, supra, at 364. In Ivan V. v. City of New York, 407 U. S. 203, 205 (1972) (per curiam), the Court held, the rule in Winship to be retroactive, because “the major purpose of the constitutional standard of proof beyond a reasonable doubt announced in Winship was to overcome an aspect of a criminal trial that substantially impairs the truth-finding function.” Consequently, our rejection in Engle of the contention advanced today — that cause need not be shown if actual prejudice is shown — is fully applicable to constitutional claims that call into question the reliability of an adjudication of legal guilt.
However, as we also noted in Engle, “[i]n appropriate cases” 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.” 456 U. S., at 135. We remain confident that, for the most part, “victims of a fundamental miscarriage of justice will meet the *496cause-and-prejudice standard.” Ibid. But we do not pretend that this will always be true. Accordingly, we think that in an extraordinary case, where a constitutional violation has probably resulted in the conviction of one who is actually innocent, a federal habeas court may grant the writ even in the absence of a showing of cause for the procedural default.
There is an additional safeguard against miscarriages of justice in criminal cases, and one not yet recognized in state criminal trials when many of the opinions on which the concurrence relies were written. That safeguard is the right to effective assistance of counsel, which, as this Court has indicated, may in a particular case be violated by even an isolated error of counsel if that error is sufficiently egregious and prejudicial. United States v. Cronic, 466 U. S. 648, 657, n. 20 (1984). See also Strickland v. Washington, 466 U. S., at 693-696. The presence of such a safeguard may properly inform this Court’s judgment in determining “[w]hat standards should govern the exercise of the habeas court’s equitable discretion” with respect to procedurally defaulted claims, Reed v. Ross, 468 U. S., at 9. The ability to raise ineffective assistance claims based in whole or in part on counsel’s procedural defaults substantially undercuts any predictions of un-remedied manifest injustices. We therefore remain of the view that adherence to the cause and prejudice test “in the conjunctive,” Engle, supra, at 134, n. 43, will not prevent federal habeas courts from ensuring the “fundamental fairness [that] is the central concern of the writ of habeas corpus.” Strickland v. Washington, supra, at 697.
The cause and prejudice test may lack a perfect historical pedigree. But the Court acknowledged as much in Wainwright v. Sykes, noting its “historic willingness to overturn or modify its earlier views of the scope of the writ, even where the statutory language authorizing judicial action has remained unchanged.” 433 U. S., at 81. The cause and prejudice test as interpreted in Engle and in our decision *497today is, we think, a sound and workable means of channeling the discretion of federal habeas courts.
I — I <1
Respondent has never alleged any external impediment that might have prevented counsel from raising his discovery claim in his petition for review, and has disavowed any claim that counsel’s performance on appeal was so deficient as to make out an ineffective assistance claim. See generally Evitts v. Lucey, 469 U. S. 387 (1985) (right to effective assistance of counsel applies on an appeal as of right). Respondent’s petition for federal habeas review of his procedurally defaulted discovery claim must therefore be dismissed for failure to establish cause for the default, unless it is determined on remand that the victim’s statements contain material that would establish respondent’s actual innocence.
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.