with whom Justice Stevens, Justice Scalía, and Justice Ginsburg join, dissenting.

The Court took this case to determine whether a vacatur is a “fact,” as that term is used in 28 U. S. C. §2255, ¶ 6(4), thus commencing the statute’s 1-year limitations period. The question divides the Courts of Appeals. Today the Court holds that the order of vacatur is the fact that begins the limitations period. On that point, I agree. Surprisingly, however, the Court proceeds to announce a second requirement of its own design: In order to obtain relief under § 2255, ¶ 6(4), petitioner must show he used due diligence in seeking the vacatur itself. On this point, I disagree.

In my view the Court’s new rule of prevacatur diligence is inconsistent with the statutory language; is unnecessary since States are quite capable of protecting themselves against undue delay in commencing state proceedings to vacate prior judgments; introduces an imprecise and incongruous deadline into the federal criminal process; is of sufficient uncertainty that it will require further litigation before its operation is understood; and, last but not least, drains scarce defense resources away from the prisoner’s federal criminal case in some of its most critical stages. For these reasons, I submit my respectful dissent.

I

The question on which we granted certiorari is this: “When a federal court bases an enhanced sentence on a vacated state conviction, is the vacatur of the state conviction a Tact’ supporting a prisoner’s 28 U. S. C. § 2255 claim requiring reduction of the prisoner’s sentence?” Pet. for Cert. i. In a change from the position it took in the Court of Appeals, the Government in its brief to this Court and again at oral argument all but conceded that the vacatur is a fact supporting a claim. See Brief for United States 33; Tr. of Oral Arg. 13. Seeking a new rationale to imprison petitioner for an *313additional eight years on the basis of a prior Georgia conviction all of us know to be void, the Government defends the Court of Appeals’ judgment on an alternative ground: Federal law requires diligence on the part of the defendant not only in bringing the vacatur to the attention of the federal court but also in commencing state proceedings to obtain the vacatur in the first place. According to the Government, petitioner’s diligence should be measured from the time a petitioner could have obtained a vacatur, i. e., as soon as the legal basis for vacatur existed. See Brief for United States 32-34. Although the Court adopts the Government’s argument in part, it comes up with a date of its own choosing from which to measure a petitioner’s diligence.

The Court is quite correct, in my view, to hold that the state-court order of vacatur itself is the critical fact which begins the Antiterrorism and Effective Death Penalty Act of 1996’s 1-year limitations period. §101, 110 Stat. 1217. Ante, at 309. It is an accepted use of the law’s vocabulary to say that the entry or the setting aside of a judgment is a fact. Ante, at 307. An order vacating a judgment is a definite and significant fact of litigation history. So the Court is on firm ground to say a state judgment of vacatur begins the 1-year limitations period. Even aside from the textual support for petitioner’s position, our opinions in Custis v. United States, 511 U. S. 485 (1994), and Daniels v. United States, 532 U. S. 374 (2001), were decided on the understanding that Congress did not expect federal sentences to be enhanced irrespective of the validity of the state conviction relied upon for the enhancement. Ante, at 305-306. Those cases suggest that the proper procedure for reducing a federal sentence enhanced on the basis of an invalid state conviction is to seek a vacatur of a state conviction, and then proceed through federal habeas.

The Court is correct, too, to say that the whole problem of vacating state-court judgments fits rather awkwardly into the language of §2255, ¶ 6(4). Ante, at 308. That is be*314cause ¶ 6(4) is designed to address myriad claims, including post-trial factual discoveries such as violations of Brady v. Maryland, 373 U. S. 83 (1963), witness recantations, new exculpatory evidence, and the like. Having gone this far, the Court in my view should simply accept that § 2255, ¶ 6(4), is not a particularly good fit with the vacatur problem.

The Court, however, does not accept the consequence of its own correct determination. Instead it finds a need to make the words “discovery” and “due diligence” more applicable to the instance of vacatur. Hence it adopts the second requirement: “[W]e also hold that the statute allows the fact of the state-court order to set the 1-year period running only if the petitioner has shown due diligence in seeking the order.” Ante, at 302. This added condition cannot be found in the statute’s design or in its text. It creates, furthermore, its own set of problems. Section 2255, ¶ 6(4), neither requires nor accommodates the Court’s federal rule of diligence respecting state-court proceedings.

II

The 1-year period begins from the date on which the facts supporting the claim or claims presented could have been discovered through the exercise of due diligence.” 28 U. S. C. § 2255, ¶ 6(4). As the Court agrees that vacatur is the fact which begins the 1-year period,' it would seem to follow that the diligence requirement pertains to presenting the fact of vacatur to the federal court. A petitioner cannot discover the vacatur until it issues. If the State has allowed the vacatur subject to its own rules respecting timely motions or applications and if petitioner has acted diligently in discovering entry of that vacatur, the proper conclusion is that he may bring a § 2255 petition within one year of obtaining the vacatur, or one year of reasonably discovering it.

The only way the majority’s construction can fit the statute is if the controlling fact is the circumstance giving rise to the vacatur, not the vacatur itself. Yet the majority re*315sists that proposition, for it measures the 1-year period from the date the vacatur is ordered. Ante, at 809.

The majority rejects petitioner’s proposed construction of the “discovered through the exercise of due diligence” language, which I would adopt, for two reasons. First, the Court observes it is “strange to say that an order vacating a conviction has been ‘discovered,’ . . . and stranger still to speak about the date on which it could have been discovered with due diligence, when the fact happens to be the outcome of a proceeding in which the § 2255 petitioner was the moving party.” Ante, at 307. By bringing vacatur proceedings, petitioner himself causes the factual event to occur, and his discovery of it is “virtually guaranteed.” Ibid. The Court is concerned that the due diligence language does barely any work under petitioner’s interpretation because the language is too easily satisfied.

Though I agree it is a bit awkward, in my view it is well within the realm of reasonable statutory construction to apply the term “discover” to an order vacating a. conviction. The ordinary meaning of the term “discovery,” after all, is “the act, process, or an instance of gaining knowledge of or ascertaining the existence of something previously unknown or unrecognized.” Webster’s Third New International Dictionary 647 (1993). See also Black’s Law Dictionary 465 (6th ed. 1990) (“[t]o get first sight or knowledge of”). There may be instances when there is a mistake in recording or entering the vacatur, or when it is not clear that the order in fact includes that relief, or when a prisoner’s transfer or extradition reasonably causes the prisoner to learn of the order in some uncommon way. In these instances, admittedly infrequent, the word “discover” makes perfect sense. True, the due diligence language does not do much work when a petitioner receives prompt notice in the ordinary course. As explained, however, § 2255, ¶ 6(4), is designed to cover various circumstances, and many other types of claims. Ante, at 310.

*316To bolster its prevacatur diligence requirement, the Court elects to resolve a case not before it, i e., a hypothetical involving DNA testing. Ibid. Quite apart from the impropriety of deciding an important question not remotely presented in the case, the Court’s resolution of its hypothetical is, in my view, far from self-evident. It has little to do, moreover, with the question of vacatur of a state-court judgment. We have a special obligation to the federal system to respect state-court judgments. Rather than imposing a federal rule of diligence on top of existing state-court rules for determining when a vacatur motion should be made, I would treat the critical fact as the date on which the state court orders vacatur. That, after all, is the time when the grounds for the claim to be made in federal court (the claim that an enhancement was improper) have become established under conventional principles commanding respect for state judgments, or allowing them to be set aside.

The second reason the majority rejects Johnson’s position is because it is troubled by the prospect that a petitioner “might wait a long time before raising any question about a predicate conviction . . . .” Ante, at 307. Even if this concern were a sufficient basis for adding the majority’s preva-catur diligence requirement to the statute and creating a two-tier diligence structure, the concern is overstated. In most instances, States can, and do, impose diligence by limiting the time for requesting a vacatur of a prior state conviction. It was represented at oral argument that all but about six States impose a limitation by statute or laches. Tr. of Oral Arg. 10. Even in those six States, furthermore, it is not clear that equitable defenses would not apply. Id., at 17-18.

Any States that do not impose time limitations are free to do so if deemed necessary to protect the integrity of their own judgments, so a federal time limit is not required. This is illustrated by the instant case. When Johnson sought state relief, Georgia imposed no limitation on a petitioner’s *317ability to obtain a vacatur. Ante, at 306, n. 5. Since then, however, Georgia has enacted a 4-year limitations period for proceeding to obtain a vacatur. The majority’s apparent concern that, absent its interpretation of § 2255, ¶ 6(4), petitioners have some incentive to delay proceedings to vacate a conviction seems quite unfounded.

The majority’s construction, furthermore, can allow for the same delay it seeks to avoid. After all, the Court holds that the due diligence requirement is triggered only by a federal judgment. Consider a simple hypothetical. Suppose that a petitioner suffers a state conviction in 1980, and, despite learning in 1985 that his conviction is constitutionally infirm, does nothing. Suppose further he is sentenced for a federal crime in 2000. Under the majority’s view, the petitioner’s obligation to question his state conviction is not triggered until 2000, a full 15 years after he knew the basis for vacatur. Despite the adaptation it makes to §2255, ¶ 6(4), the majority has failed to create an incentive for petitioner to act promptly in instituting state proceedings. The incentive exists under state law, and the Court does not need to supplement it.

The error of the majority’s position is further revealed by its selection of what I consider to be an incorrect date for triggering the prevacatur diligence requirement. It holds that the triggering event is set at the date of petitioner’s federal judgment. Ante, at 310 (setting November 29,1994, the date of judgment, as the date triggering the diligence requirement).

This rule of the Court’s own contrivance is adopted, in my respectful submission, without full appreciation for the dynamic of the criminal process and its demands on counsel. Assuming for the moment that some event in the federal court should start the time period for pursuing state relief, surely the entry of judgment is ill chosen. This means the judgment is a mandatory beginning point for collateral proceedings to correct a judgment and sentence not yet final.

*318If the Court wants to invent its own rule and use an event in the federal criminal proceeding to commence a limitations period (and I disagree with both propositions), the date the judgment becomes final, not the date of judgment in the trial court, is the proper point of beginning.

The law, and the decisions of this Court, put extraordinary demands on defense counsel. Immediately after a judgment, defense counsel must concentrate on ensuring that evidence of trial misconduct does not disappear and that grounds for appeal are preserved and presented. Today the Court says defense counsel must divert scarce resources from these heavy responsibilities to commence collateral proceedings to attack state convictions.

In this case seven different convictions in Georgia may have been relevant. In other cases convictions that might enhance have been entered in different States. See, e.g., Custis, 511 U. S., at 487. It is most troubling for a Court that insists on high standards of performance for defense counsel now to instruct that collateral proceedings must be commenced in one or more States during the critical time immediately after judgment and before appeal.

If the Court is to insist upon its own second tier of diligence, the dynamics of the criminal system and ordinary rules for determining when collateral proceedings become necessary should instruct us that, for federal purposes, this tier begins when the federal conviction becomes final. This also ensures that the federal court does not make demands on counsel and on state courts that are pointless if the federal conviction is overturned. Perhaps the Court rejects the date of final judgment as triggering its requirement because it adds little to the state requirements of diligence. If this surmise is correct, of course, it demonstrates that the Court should not adopt its interpretation in the first place.

Aside from diverting resources from a petitioner’s federal case, the majority’s approach creates new uncertainty, giving rise to future litigation. It leaves unsaid what standard will *319be used for measuring whether a petitioner acted promptly, forcing litigants and lawyers to scramble to state court in the hopes they satisfy the Court’s vague prevacatur diligence requirement. The Court tells us nothing about what to make of existing state standards regarding diligence. Assume a State has a 4-year limitations period for bringing a vacatur action and a petitioner acts within two years of his state conviction. Do we look to state law as a benchmark for what should be presumed to be diligent? The murkiness of the Court’s new rule will set in motion satellite litigation on this and related points.

In lieu of adopting an interpretation that creates more problems than it avoids, I would hold that the order vacating a prior state conviction is the fact supporting a § 2255 claim, and the statute is satisfied if the § 2255 proceeding is commenced within one year of its entry, unless the petitioner shows it was not reasonably discovered until later in which case that date will control when the statute begins to run. For these reasons, I would reverse the judgment of the Court of Appeals.

delivered the opinion of the Court.

The question here is when the 1-year statute of limitations in 28 U. S. C. § 2255, ¶ 6(4), begins to run in a case of a prisoner’s collateral attack on his federal sentence on the ground that a state conviction used to enhance that sentence has since been vacated. We hold that the period begins when a petitioner receives notice of the order vacating the prior conviction, provided that he has sought it with due diligence in state court, after entry of judgment in the federal case with the enhanced sentence.

I — (

In 1994, petitioner Robert Johnson, Jr., was indicted for distributing cocaine base and related conspiracy. Following his guilty plea to a single count of distribution in violation of 21 U. S. C. § 841(a)(1) and 18 U. S. C. § 2, the presentence investigation report recommended that Johnson receive an enhanced sentence as a career offender under §4B1.1 of the Federal Sentencing Guidelines, owing to his two 1989 convictions by the State of Georgia for distributing cocaine. Without elaboration, Johnson filed an objection to the recommendation, which he withdrew at the sentencing hearing. The District Court imposed the enhancement and entered judgment on November 29, 1994.

On appeal, Johnson argued for the first time that he should not have been sentenced as a career offender because one of *299his Georgia convictions was invalid.1 The Court of Appeals for the Eleventh Circuit affirmed the sentence, finding that in the trial court Johnson had raised no objection to the validity of his prior convictions and that the judge’s career offender findings were not clearly erroneous. United States v. Johnson, No. 94-9402 (Dec. 22, 1995) (per curiam), App. 7. In a footnote, the Court of Appeals

“note[d] in passing that, should appellant obtain at some future date the vacation of the state court conviction in question because [it was] obtained in violation of his constitutional rights, he could petition the district court under 28 U. S. C. § 2255 for the relief he now asks us to provide.” Id., at 8, n. 1.

We denied certiorari. Johnson v. United States, 517 U. S. 1162 (1996).

Two days later, the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) went into effect, imposing, among other things, a 1-year period of limitations on motions by prisoners seeking to modify their federal sentences:

“The limitation period shall run from the latest of—

“(1) the date on which the judgment of conviction becomes final;

“(2) the date on which the impediment to making a motion created by governmental action in violation of *300the Constitution or laws of the United States is removed, if the movant was prevented from making a motion by such governmental action;

“(3) the date on which the right asserted was initially recognized by the Supreme Court, if that right has been newly recognized by the Supreme Court and made retroactively applicable to cases on collateral review; or

“(4) the date on which the facts supporting the claim or claims presented could have been discovered through the exercise of due diligence.” 28 U. S. C. §2255, ¶6.

A fifth option supplied uniformly by the Courts of Appeals gave prisoners whose convictions became final before AEDPA a 1-year grace period running from the new statute’s effective date. Duncan v. Walker, 533 U. S. 167, 183, n. 1 (2001) (Stevens, J., concurring in part and concurring in judgment) (collecting cases).

On April 25, 1997, one year and three days after his pre-AEDPA federal conviction became final and just after the 1-year grace period expired, Johnson pro se filed a motion in the District Court for a 60-day extension of time to attack his federal sentence under § 2255.2 Finding the AEDPA period expired, the District Court denied the motion, though it added that denial was without prejudice to Johnson’s right to file a § 2255 motion claiming any alternative limitation period under the statute.

On February 6,1998, Johnson petitioned for writ of habeas corpus in the Superior Court of Wayne County, Georgia, claiming the invalidity of his guilty pleas in seven cases between 1983 and 1993 because he had not knowingly, intelligently, and voluntarily waived his right to counsel. One of the seven pleas Johnson challenged was the basis for one of *301the 1989 convictions on which the District Court had rested the career offender enhancement of Johnson’s federal sentence.3 The State of Georgia denied Johnson’s allegations, but filed no hearing transcripts. The Superior Court found that the records did “not show an affirmative waiver of [Johnson’s] right to an attorney” in any of the cases, App. 10, and entered an order of vacatur, that all seven convictions be reversed, ibid.

Just over three months later, Johnson was back in the Federal District Court pro se with a motion under § 2255 to vacate the enhanced federal sentence following the vacatur of one of its predicate state convictions. He claimed his motion was timely because the order vacating the state judgment was “new evidence” not previously discoverable, and so the trigger of a renewed limitation period. The Magistrate Judge took Johnson to be relying on the discovery of “facts supporting the claim” addressed in § 2255, ¶ 6(4), but still recommended denial of the motion for failure on Johnson’s part to exercise the “due diligence” required by that provision. Id., at 15-17. Although Johnson objected that lack of education excused him from acting more promptly, and that he had filed the state petition as soon as he could get help from an inmate law clerk, the District Court denied the § 2255 motion as untimely. In the court’s view, the applicable limitation was the 1-year grace period that was over in April 1997, which Johnson had done nothing to toll in the 21 months he waited after his conviction became final before filing his state habeas petition. Id., at 19.

A divided panel of the Court of Appeals affirmed. 340 F. 3d 1219 (CA11 2003). The majority reasoned that the state-court order vacating the prior state conviction was not *302a “‘fact’” under the fourth paragraph of the §2255 limitation rule. Id., at 1223. In the majority’s view, the state-court order was properly classified as a “legal proposition” or a “court action obtained at the behest of a federal prisoner, not ‘discovered’ by him.” Ibid. Because the fourth paragraph of the limitation rule was therefore of no avail to Johnson, the Court of Appeals majority agreed with the District Court that the time for filing expired in 1997, at the end of the 1-year grace period. Id., at 1226. The majority also agreed that Johnson had no equitable claim to toll the running of the 1-year period because he had waited too long before going back to the state court. Id., at 1226-1228. Judge Roney dissented, arguing that the state court’s order was a “fact” supporting Johnson’s § 2255 motion, a fact not discoverable prior to the order’s issuance. Id., at 1228-1229. Over one dissent, rehearing en banc was denied. 353 F. 3d 1328 (2003).

We granted certiorari, 542 U. S. 965 (2004), to resolve a disagreement among the Courts of Appeals as to whether vacatur of a prior state conviction used to enhance a federal sentence can start the 1-year limitation period under the fourth alternative of the §2255 rule. Compare 340 F. 3d 1219 (case below) (vacatur not a trigger); Brackett v. United States, 270 F. 3d 60 (CA1 2001) (same), with United States v. Gadsen, 332 F. 3d 224 (CA4 2003) (vacatur a fact not previously discoverable giving rise to a new 1-year period).

We agree with Johnson that the state-court vacatur is a matter of fact for purposes of the limitation rule in the fourth paragraph. But we also hold that the statute allows the fact of the state-court order to set the 1-year period running only if the petitioner has shown due diligence in seeking the order. Applying that qualification, we affirm.

hH l-H

The Government shares Johnson s preliminary assumption that if he filed his §2255 motion in time, he is entitled to *303federal resentencing now that the State has vacated one of the judgments supporting his enhanced sentence. Neither the enhancement provision of the Sentencing Guidelines applied here, nor the mandatory enhancement under the Armed Career Criminal Act (ACCA), 18 U. S. C. § 924(e), has been read to mean that the validity of a prior conviction supporting an enhanced federal sentence is beyond challenge. Cf. Lewis v. United States, 445 U. S. 55 (1980) (validity of prior conviction irrelevant under federal statute prohibiting possession of a firearm by a felon). Our cases applying these provisions assume the contrary, that a defendant given a sentence enhanced for a prior conviction is entitled to a reduction if the earlier conviction is vacated. Custis v. United States, 511 U. S. 485 (1994); Daniels v. United States, 532 U. S. 374 (2001).

Such was the premise in Custis v. United States, supra, even though we held that the ACCA generally created no opportunity to attack a prior state conviction collaterally at a federal sentencing proceeding, 511 U. S., at 490, and that the Constitution demands no more, id., at 496-497. We thought that Congress had not meant to make it so easy to challenge final judgments that every occasion to enhance a sentence for recidivism would turn a federal sentencing court into a forum for difficult and time-consuming reexaminations of stale state proceedings. Ibid. We recognized only one exception to this rule that collateral attacks were off-limits, and that was for challenges to state convictions allegedly obtained in violation of the right to appointed counsel, an exception we thought necessary to avoid undermining Gideon v. Wainwright, 372 U. S. 335 (1963). Custis v. United States, 511 U. S., at 494-496. As to challenges falling outside of that exception, we pointed out that a defendant who successfully attacked his state conviction in state court or on federal habeas review could then “apply for reopening of any federal sentence enhanced by the state sentences.” Id., at 497.

*304Daniels v. United States, supra, extended Custis to hold, subject to the same exception for Gideon claims, that a federal prisoner may not attack a predicate state conviction through a §2255 motion challenging an enhanced federal sentence, 532 U. S., at 376, and again we stressed considerations of administration and finality, id., at 378-380. Again, too, we acknowledged that a prisoner could proceed under §2255 after successful review of the prior state conviction on federal habeas under §2254 or favorable resort to any postconviction process available under state law, id., at 381. We simply added that if the prior conviction was no longer open to direct or collateral attack in its own right, the federal prisoner could do nothing more about his sentence enhancement. Id., at 382.4

This case presents the distinct issue of how soon a prisoner, successful in his state proceeding, must challenge the federal sentence under § 2255. The resolution turns on understanding what “facts” affecting an enhanced sentence could most sensibly fall within that term as used in the fourth paragraph of the §2255 limitation provision, under which the one year runs from “the date on which the facts supporting the claim or claims presented could have been discovered through the exercise of due diligence.” Johnson says that the order vacating his prior conviction is the factual matter supporting his § 2255 claim, discovery of which triggers the refreshed 1-year period. The Court of Appeals majority said no because it understood a legally operative order of vacatur to be a mandate of law or a consequence of applying law, and therefore distinct from a matter of “ 'fact’ ” *305as Congress used the term in §2255. 340 F. 3d, at 1223. The United States does not endorse that law-fact distinction, but argues that the facts supporting Johnson’s § 2255 claim, for purposes of the fourth paragraph, are the facts on which he based his challenge to the validity of his state convictions.

We think none of these positions is sound, at least in its entirety. As for the Government’s proposed reading, certainly it is true that the circumstances rendering the underlying predicate conviction invalid are ultimate subjects of fact supporting the §2255 claim, in the sense that proof of those facts (or the government’s failure to negate them) is necessary to vacate the prior state conviction and eliminate the ground for the federal enhancement. But this is not enough to fit the Government’s position comfortably into paragraph.four. The text of §2255, ¶6(4), clearly links the running of the limitation period to the discovery of the “facts supporting the claim or claims presented,” but on the Government’s view, the statute of limitations may begin to run (and may even expire) before the § 2255 claim and its necessary predicate even exist. Prior to the federal conviction, a petitioner has no §2255 claim because he has no enhanced federal sentence to challenge; and prior to the state vacatur, which Daniels makes a necessary condition for relief in most cases, a petitioner cannot obtain relief under § 2255. Cf. Bay Area Laundry and Dry Cleaning Pension Trust Fund v. Ferbar Corp., of Cal., 522 U. S. 192, 195 (1997) (statutes of limitations ordinarily do not begin to run until a plaintiff’s complete cause of action has accrued). Hence, it is highly doubtful that in §2255 challenges to enhanced sentences Congress would have meant to start the period running under paragraph four on the discoverability date of facts that may have no significance under federal law for years to come and that cannot by themselves be the basis of a § 2255 claim, Daniels v. United States, supra, at 376.

There are further reasons against applying the fourth paragraph as the Government would. Congress does not *306appear to have adopted a policy of enhancing federal sentences regardless of the validity of state convictions relied on for the enhancement. Custis and Daniels were decided on just the contrary, and unchallenged, understanding; it would certainly push the limits of coherence for the Court now to apply the fourth paragraph in a way that would practically close the door to relief that each of those cases specifically left open.5 Nor is there any reason to think Congress meant the limitation period to run earlier for the sake of preserving finality of state convictions; States are capable of providing their own limitations periods (and most of them would have barred Johnson’s challenge).6

Johnson’s argument improves on the Government’s proposal by pegging the limitation period to notice of the state order eliminating the predicate required for enhancement, which is almost always necessary and always sufficient for relief. We do not find his proposal vulnerable to the point made by the majority of the Court of Appeals, that an order vacating a conviction is legally expressive or operative language that may not be treated as a matter of fact within the meaning of the statute. We commonly speak of the “fact of *307a prior conviction,” e. g., Apprendi v. New Jersey, 530 U. S. 466, 490 (2000), and an order vacating a predicate conviction is spoken of as a fact just as sensibly as the order entering it. In either case, a claim of such a fact is subject to proof or disproof like any other factual issue.

But Johnson’s take on the statute carries anomalies of its own, one minor, one more serious. It is strange to say that an order vacating a conviction has been “discovered,” the term used .by paragraph four, and stranger still to speak about the date on which it could have been discovered with due diligence, when the fact happens to be the outcome of a proceeding in which the §2255 petitioner was the moving party. By bringing that proceeding, the petitioner causes the factual event to occur, after all, and unless his mail goes astray his prompt discovery of the crucial fact is virtually guaranteed through official notice.

A more serious problem is Johnson’s position that his § 2255 petition is timely under paragraph four as long as he brings it within a year of learning he succeeded in attacking the prior conviction, no matter how long he may have slumbered before starting the successful proceeding. If Johnson were right about this, a petitioner might wait a long time before raising any question about a predicate conviction, as this very case demonstrates. Of course it may well be that Johnson took his time because his basic sentence had years to run before the period of enhancement began. But letting a petitioner wait for as long as the enhancement makes no difference to his actual imprisonment, while the predicate conviction grows increasingly stale and the federal outcome is subject to question, is certainly at odds with the provision in paragraph four that the one year starts running when the operative fact “could have been discovered through the exercise of due diligence.” And by maximizing the time that judgments are open to question, a rule allowing that kind of delay would thwart one of AEDPA’s principal purposes, *308Duncan v. Walker, 533 U. S., at 179; Woodford v. Garceau, 538 U. S. 202, 206 (2003), a purpose that was also central to our decisions in Custis and Daniels, see supra, at 303-304.

We think neither anomaly is serious enough, however, to justify rejecting Johnson’s basic argument that notice of the order vacating the predicate conviction is the event that starts the one year running. Our job here is to find a sensible way to apply paragraph four when the truth is that with Daniels not yet on the books AEDPA’s drafters probably never thought about the situation we face here. Of course it is peculiar to speak of “discovering” the fact of the very eventuality the petitioner himself has brought about, but when that fact is necessary to the § 2255 claim, and treating notice of it as the trigger produces a more reasonable scheme than the alternatives, the scheme should be reconciled with the statutory language if it can be. And here the fit is painless, if short on style.

While it sounds odd to speak of discovering a fact one has generated, a petitioner does not generate the fact of vacatur all by himself. He does, after all, have to learn of the court’s response in the state proceeding, and receiving notice of success can surely qualify as a kind of discovery falling within the statutory language.

That leaves us with the question of how to implement the statutory mandate that a petitioner act with due diligence in discovering the crucial fact of the vacatur order that he himself seeks. The answer is that diligence can be shown by prompt action on the part of the petitioner as soon as he is in a position to realize that he has an interest in challenging the prior conviction with its potential to enhance the later sentence. The important thing is to identify a particular time when the course of the later federal prosecution clearly shows that diligence is in order. That might be the date the federal indictment is disclosed, the date of judgment, or the date of finality after direct appeal. Picking the first date would require the quickest response and serve finality best, *309but it would produce some collateral litigation that federal acquittals would prove to have been needless, and it shares the same disconnection from the existence of a §2255 claim as the Government’s view of the relevant “facts,” see supra, at 305-306. If we picked the third date, collateral litigation would be minimized, but finality would come late. This shapes up as a case for choosing the bowl of porridge between the one too hot and the one too cold, and settling on the date of judgment as the moment to activate due diligence seems best to reflect the statutory text and its underlying concerns. After the entry of judgment, the subject of the § 2255 claim has come into being, the significance of inaction is clear, and very little litigation would be wasted, since most challenged federal convictions are in fact sustained.

The dissent, like Johnson, would dispense with any due diligence requirement in seeking the state vacatur order itself, on the ground that the States can impose their own limitations periods on state collateral attacks, as most States do, post, at 316 (opinion of Kennedy, J.). But the United States has an interest in the finality of sentences imposed by its own courts; §2255 is, after all, concerned directly with federal cases. As to those federal cases, due diligence is not a “requirement of [our] own design,” post, at 312, but an explicit demand in the text of § 2255, ¶ 6(4), one that reflects AEDPA’s core purposes, supra, at 307-308. The requirement of due diligence must therefore demand something more than the dissent’s willingness to accept no diligence at all, if the predicate conviction occurred in a State that itself imposes no limit of time for collaterally attacking its convictions.7

*310The dissent suggests that due diligence is satisfied by-prompt discovery of the existence of the order vacating the state conviction. Post, at 314. Where one “discovers” a fact that one has helped to generate, however, supra, at 308-309, whether it be the result of a court proceeding or of some other process begun at the petitioner’s behest, it does not strain logic to treat required diligence in the “discovery” of that fact as entailing diligence in the steps necessary for the existence of that fact. To see why this is so, one need only consider a more commonplace use of the paragraph four limitation rule. When a petitioner bases his § 2255 claim on the result of a DNA test, it is the result of the test that is the “fac[t] supporting the claim” in the § 2255 motion, and the 1-year limitation period therefore begins to run from the date the test result is “discovered.” Yet unless it is to be read out of the statute, the due diligence requirement would say that the test result only triggers a new 1-year period if the petitioner began the testing process with reasonable promptness once the DNA sample and testing technology were available. Under the dissent’s view, however, the petitioner could wait untold years (perhaps until the death of a key prosecution witness) before calling for the DNA test, yet once he “discovered” the result of that test, he would get the benefit of a rejuvenated 1-year period regardless of his lengthy delay. Such a result simply cannot be squared with the statute’s plain text and purpose.

We accordingly apply the fourth paragraph in the situation before us by holding that from November 29, 1994, the date the District Court entered judgment in his federal case, Johnson was obliged to act diligently to obtain the state-court order vacating his predicate conviction. Had he done so, the 1-year limitation period would have run from the date he received notice of that vacatur.8

*311II

Although Johnson knew that his conviction subjected him to the career offender enhancement, he failed to attack the predicate for enhancement by filing his state habeas petition until February 1998, more than three years after entry of judgment in the federal case. Indeed, even if we moved the burden of diligence ahead to the date of finality of the federal conviction or to AEDPA’s effective date two days later, Johnson would still have delayed unreasonably, having waited over 21 months. Johnson has offered no explanation for this delay, beyond observing that he was acting pro se and lacked the sophistication to understand the procedures. But we have never accepted pro se representation alone or procedural ignorance as an excuse for prolonged inattention when a statute’s clear policy calls for promptness, and on this record we think Johnson fell far short of reasonable diligence in challenging the state conviction. Since there is every reason to believe that prompt action would have produced a state vacatur order well over a year before he filed his § 2255 petition, the fourth paragraph of §2255 is unavailable, and Johnson does not suggest that his motion was timely under any other provision.

We accordingly affirm the judgment of the Court of Appeals.

It is so ordered.