Justice Breyer, with whom Justice Stevens joins, concurring in the judgment.

I believe the Court of Appeals correctly decided the basic question: “Whether a federal habeas court is barred from considering an ineffective-assistance-of-counsel claim as ’cause’ for the procedural default of another claim when the ineffective-assistance claim is itself procedurally defaulted.” The question’s phrasing itself reveals my basic concern. Although the question, like the majority’s opinion, is written with clarity, few lawyers, let alone unrepresented state prisoners, will readily understand it. The reason lies in the complexity of this Court’s habeas corpus jurisprudence—a complexity that in practice can deny the fundamental constitutional protection that habeas corpus seeks to assure. Today’s decision unnecessarily adds to that complexity and cannot be reconciled with our consistent recognition that the determination of “cause” is a matter for the federal habeas judge.

To explain why this is so, and at the risk of oversimplification, I must reiterate certain elementary ground rules. A federal judge may issue a writ of habeas corpus freeing a state prisoner, if the prisoner is “in custody in violation of the Constitution or laws or treaties of the United States.” 28 U. S. C. §2254(a). However, the judge may not issue the writ if an adequate and independent state-law ground justifies the prisoner’s detention, regardless of the federal claim. See Wainwright v. Sykes, 433 U. S. 72, 81–88 (1977). One “state ground” often asserted as an adequate, independent basis for holding a state prisoner in custody is a state-law “procedural default,” such as the prisoner's failure to raise his federal claim at the proper time. However, under certain conditions the State’s assertion of such a ground is not “adequate” (and consequently does not bar assertion of the federal-law claim). There are three situations in which an otherwise valid state ground will not bar federal claims: (1) where failure to consider a prisoner’s claims will result in a “fundamental miscarriage of justice,” Coleman v. Thompson, 501 U. S. 722, 750 (1991); (2) where the state procedural rule was not “ ‘firmly established and regularly followed,’ ” Ford v. Georgia, 498 U. S. 411, 423–424 (1991); James v. Kentucky, 466 U. S. 341, 348–349 (1984); and (3) where the prisoner had good “cause” for not following the state procedural rule and was “prejudice[d]” by not having done so, Sykes, supra, at 87.

Ordinarily, a federal habeas judge, while looking to state law to determine the potential existence of a procedural ground that might bar consideration of the prisoner’s federal claim, decides whether such a ground is adequate as a matter of federal law. See Ford, supra; James, supra; Coleman, supra. Thus the Court has applied federal standards to determine whether there has been a “fundamental miscarriage of justice.” See, e.g., Schlup v. Delo, 513 U. S. 298, 314–317 (1995). And the Court has also looked to state practice to determine the factual circumstances surrounding the application of a state procedural rule, while determining as a matter of federal law whether that rule is “firmly established [and] regularly followed.” Ford, supra, at 424–425. Federal habeas courts would normally determine whether “cause and prejudice” excuse a “procedural default” in the same manner. Murray v. Carrier, 477 U. S. 478, 489 (1986) (“[T]he question of cause” is “a question of federal law”).

If I could stop here, the rules would be complicated, but still comprehensible. The federal habeas judge would look to state law and state practice to determine the facts and circumstances surrounding a state procedural rule that the State claims is an “adequate and independent state ground.” However, the federal judge would determine the adequacy of that “state ground” as a matter of federal law.

Unfortunately, the rules have become even more complex. In Carrier, the Court considered a prisoner’s contention that he had “cause” for failing to follow a state procedural rule—a rule that would have barred his federal claim. The “cause,” in the prisoner’s view, was that his lawyer (who had failed to follow the state procedural rule) had performed inadequately. This Court determined, as a matter of federal law, that only a performance so inadequate that it violated the defendant’s Sixth Amendment right to effective assistance of counsel could amount to “cause” sufficient to overcome a “procedural default.” Id., at 488–489. That being so, the Court reasoned, the prisoner should have to exhaust the ineffectiveness claim in state court. The Court wrote:

“[I]f a petitioner could raise his ineffective assistance claim for the first time on federal habeas 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.” Id., at 489.

And today the Court holds not only that the prisoner must exhaust this claim by presenting it to the state courts, but also that his failure to do so properly, i.e., a failure to comply with the State’s rules for doing so, bars that prisoner from ever asserting that claim as a “cause” for not having complied with state procedural rules.

The opinion in Carrier raises a special kind of “exhaustion” problem. The Court considered a type of “cause” (“ineffective assistance”) for not following the state procedural rule that happened itself independently to constitute a violation of the Federal Constitution. After all, were the prisoner to prove his claim (i.e., show “ineffective assistance”), the State might want to take action first. Ordinary exhaustion rules assure States an initial opportunity to pass upon claims of violation of the Federal Constitution. Why should a State not have a similar opportunity in this situation? As the Carrier Court pointed out, it would be “anomalous” for a federal habeas court to “adjudicat[e] an unexhausted constitutional claim for which state court review might still be available.” Ibid.

The anomaly disappears, however, once the prisoner has exhausted his “ineffective-assistance” claim (which appeared in the guise of a “cause”). And there is no other anomaly that requires the majority’s result. Once a claim of ineffective assistance of counsel has been exhausted—either through presentation in the state courts or through procedural default—there is no difference between that claim and any other claim of “cause” for the prisoner’s original procedural default. The federal habeas court is no longer in the “anomalous position” of considering as cause an independent claim that might yet be considered by the state courts, for there is no longer any possibility that the state courts will consider the claim. There is thus no more reason to hold that procedural default of an ineffective-assistance claim bars the prisoner from raising thatineffective-assistance claim as a “cause” (excusing a different procedural default asserted as a bar to a basic constitutional claim) than there is to bar any other claim of “cause” on grounds of procedural default. The majority creates an anomaly; it does not cure one.

The added complexity resulting from the Court’s opinion is obvious. Consider a prisoner who wants to assert a federal constitutional claim (call it FCC). Suppose the State asserts as a claimed “adequate and independent state ground” the prisoner’s failure to raise the matter on his first state-court appeal. Suppose further that the prisoner replies by alleging that he had “cause” for not raising the matter on appeal (call it C). After Carrier, if that alleged “cause” (C) consists of the claim “my attorney was constitutionally ineffective,” the prisoner must have exhausted C in the state courts first. And after today, if he did not follow state rules for presenting C to the state courts, he will have lost his basic claim, FCC, forever. But, I overstate. According to the opinion of the Court, he will not necessarily have lost FCC forever if he had “cause” for not having followed those state rules (i.e., the rules for determining the existence of “cause” for not having followed the state rules governing the basic claim, FCC) (call this “cause” C*). Ante, at 6–7. The prisoner could therefore still obtain relief if he could demonstrate the merits of C*, C, and FCC.

I concede that this system of rules has a certain logic, indeed an attractive power for those who like difficult puzzles. But I believe it must succumb to this question: Why should a prisoner, who may well be proceeding pro se, lose his basic claim because he runs afoul of state procedural rules governing the presentation to state courts of the “cause” for his not having followed state procedural rules for the presentation of his basic federal claim? And, in particular, why should that special default rule apply when the “cause” at issue is an “ineffective-assistance-of-counsel” claim, but not when it is any of the many other “causes” or circumstances that might excuse a failure to comply with state rules? I can find no satisfactory answer to these questions.

I agree with the majority, however, that this case must be returned to the Court of Appeals. Although the prisoner’s “ineffective-assistance” claim is not barred, he still must prove that the “assistance” he received was “ineffective” (or some other “cause”). And, if he does so, he still must prove his basic claim that his trial violated the Federal Constitution—all before he can secure habeas relief. I would remand for consideration of these matters.

For these reasons, I concur in the judgment.