with whom The Chief Justice, Justice O'Connor, and Justice Thomas join, dissenting.
Today’s opinion permits review where Congress, with unmistakable clarity, has denied it. To reach this result, the Court ignores the obvious intent of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), Pub. L. 104-132, 110 Stat. 1214, distorts the meaning of our own jurisdictional statute, 28 U. S. C. § 1254(1), and overrules a 53-year-old precedent, House v. Mayo, 324 U. S. 42 (1945) (per curiam). I respectfully dissent.
I
This Court’s jurisdiction under 28 TJ. S. C. § 1254(1) is limited to “[cjases in the courts of appeals.” Section 102 of AEDPA provides that “[ujnless a circuit justice or judge issues a certificate of appealability, an appeal may not be taken to the court of appeals from . . . the final order in a habeas eorpus proceeding under section 2255,” that is, a district court habeas proceeding challenging federal custody. Petitioner, who is challenging federal custody under 28 U. S. C. §2255, did not obtain a certificate of appealability (COA). By the plain language of AEDPA, his appeal “from” the district court’s “final order” “may not be taken to the court of appeals.” Because it could not be taken to the Court of Appeals, it quite obviously was never in the Court of Appeals; and because it was never in the Court of Appeals, we lack jurisdiction under § 1254(1) to entertain it.
We have already squarely and explicitly endorsed this straightforward interpretation. In House v. Mayo, 324 U. S., at 44, involving the predecessors to §§1254(1) and *2552253(c)(1), the statutorily required certificate was called a “certificate of probable cause” rather than a certificate of ap-pealability, but the effect of failure to obtain it was precisely the same: The ease could not proceed to the court of appeals. On an attempt to obtain review of denial of the certificate in this Court, we held that since petitioner’s “case was never 'in’ the court of appeals, for want of a certificate,” we lacked jurisdiction under § 1254(1). Ibid.
The Court concedes that House is squarely on point but opts to overrule it because its “conclusion was erroneous,” ante, at 251. The Court does not dispute that petitioner’s §2255 action was never in the Court of Appeals; its overruling of House is instead based on the proposition that petitioner’s request for a COA is, in and of itself, a “ease” within the meaning of § 1254(1), see ante, at 241-242, 246-249, and that that case was “in” the Court of Appeals and hence can be reviewed here, ante, at 241-246. Most of the Court’s analysis is expended in the effort to establish that petitioner made his request for a COA to the Court of Appeals as such, rather than to the circuit judges in their individual capacity, ibid. Even that effort is unsuccessful, since it comes up against the pellucid language of AEDPA to the contrary. Section 102 does not permit application for a COA to a court of appeals; it states that the application must be made to a “circuit justice or judge.” That this means precisely what it says is underscored by § 103 of AEDPA, which amends Rule 22 of the Federal Rules of Appellate Procedure: “If [a COA] request is addressed to the eourt of appeals, it shall be deemed addressed to the judges thereof and shall be considered by a circuit judge or judges as the eourt deems appropriate.” As though drafted in anticipatory refutation of the Court’s eountertextual holding today, the Advisory Committee’s Notes on Rule 22 explicitly state that “28 U. S. C. §2253 does not authorize the court of appeals as a eourt to grant a certificate of probable cause.” 28 U. S. C. App., p. 609.
*256Proclaiming the request for a COA to be "in” the Court of Appeals is the most obvious of the Court’s statutory distortions, but not the one with the most serious collateral consequences. The latter award goes to the Court’s virtually un-analyzed pronouncement (also essential to its holding) that the request for a COA was itself a "case” within the meaning of § 1254(1). The notion that a request pertaining to a case constitutes its own "case” for purposes of § 1254 is a jaw-dropper. To support that remarkable assertion, the Court relies upon circumstantial evidence — that the "application moved through the Eighth Circuit in the same manner as cases in general do.” Ante, at 242. Does this mean that a request for a COA would not be a "case” in those Circuits that treated it differently — that permitted it to be disposed of by a single judge as Rule 22 specifically allows? Does it mean that a motion for recusal, or a request for televised coverage, or a motion to file under seal would be a "case” if the court of appeals chose to treat it in the manner the Eighth Circuit treated the request for a COA here? Surely not.
An application for a COA, standing alone, does not have the requisite qualities of a legal “case” under any known definition. It does not assert a grievance against anyone, does not seek remedy or redress for any legal injury, and does not even require a “party” on the other side. It is nothing more than a request for permission to seek review. Petitioner’s grievance is with respondent for unlawful custody, and the remedy he seeks is release from that custody pursuant to § 2265. The request for a COA is not some separate “ease” that can subsist apart from that underlying suit; it is merely a procedural requirement that must be fulfilled before petitioner’s § 2255 action — his “case” or “cause” — can advance to the appellate court. The adversity which the Court acknowledges is needed for a “case” under § 1254, see ante, at 241, is not satisfied by the dispute between petitioner and respondent as to whether the COA should be granted— *257any more than a “ease or controversy” for purposes of initial federal-court jurisdiction is created by a dispute over venue, between parties who agree on everything else.1
As is true with most erroneous theories, a logical and consistent application of the Court’s reasoning yields strange results. If dispute over the propriety of granting a COA creates a “ease,” the denial of a COA request that has been unopposed (or, better yet, has been supported by the Government) will be unreviewable, whereas denial of a request that is vigorously opposed will be reviewed — surely an upside-down result. And the “case” concerning the COA will subsist even when the §2255 suit has been mooted by the petitioner’s release from prison. These bizarre consequences follow inevitably from the Court’s “separate- case” theory, which has been fabricated in order to achieve a result that is fundamentally at odds with the purpose of the statute. For the Court insists upon assuming, contrary to the plain import of the statute, that Congress wanted petitioner’s § 2255 action to proceed “in the ordinary course of the judicial process” and to follow the “general rule” that permits an appeal from a final district court order, ibid. If this were Congress’s wish, there would have been no need for § 102 of AEDPA. The whole point of that provision is to diverge from the ordinary course of the judicial process and to keep petitioner’s case against respondent out of the Court of Appeals unless petitioner obtains a COA. “The certificate is a screening device, helping to conserve judicial (and prosecuto-rial) resources.” Young v. United States, 124 F. 3d 794, 799 *258(CA7 1997). It is this unique screening function that distinguishes a COA from the jurisdictional issues discussed by the Court: Section 102 of AEDPA prevents petitioner’s ease from entering the Court of Appeals at all in the absence of a COA, whereas other jurisdictional determinations are made after a case is in the Court of Appeals (even if the ease is later dismissed because of jurisdictional defects), ante, at 246-249. See Rosado v. Wyman, 397 U. S. 397, 403, n. 3 (1970) (a court always has jurisdiction to determine its jurisdiction).
The Court’s only response to these arguments is that they are foreclosed by our precedent, since we decided an analogous issue in Ex parte Quirin, 317 U. S. 1 (1942). Ante, at 246. (The Court displays no appreciation of the delicious irony involved in its insistence upon hewing to an allegedly analogous decision while overruling the case directly in point, House.) Quirin held that a petition for habeas corpus constituted the institution of a suit, and that it was not necessary for the writ to issue for the matter to be considered a case or controversy. 317 U. S., at 24. Quirin relied upon our decision in Ex parte Milligan, 4 Wall. 2, 110-113 (1866), which reasoned that a petition for habeas corpus is a suit because the petitioner seeks “That remedy which the law affords him’” to recover his liberty. Id., at 113 (quoting Weston v. City Council of Charleston, 2 Pet. 449, 464 (1829)). Petitioner’s request for §2255 relief is analogous to a petition for habeas corpus, but his request for a COA is of a wholly different nature. That is no “remedy” for any harm, but a threshold procedural requirement that petitioner must meet in order to carry his §2255 suit to the appellate stage. That is why the Court in House, decided less than three years after Quirin, did not treat the application for a certificate as a separate case but did recognize the petition for habeas corpus as a ease even though it was decided without a hearing or a call for a return. 324 U. S., at 43.
I have described above why House was entirely correct, but a few words are in order concerning the inappropriate*259ness of overruling House, regardless of its virtue as an original matter. “[T]he "burden borne by the party advocating abandonment of an established precedent is greater where the Court is asked to overrule a point of statutory construction.” Patterson v. McLean Credit Union, 491 U. S. 164, 172-173 (1989); see also Illinois Brick Co. v. Illinois, 431 U. S. 720, 736 (1977); Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 406 (1932) (Brandeis, J., dissenting). The Court acknowledges this principle, but invokes cases of ours that say that stare decisis concerns are “ ‘somewhat reduced’ ” in the ease of a procedural rule. Ante, at 251. The basis for that principle, of course, is that procedural rules do not ordinarily engender detrimental reliance — and in this ease, as I shall discuss, detrimental reliance by the Congress of the United States is self-evident. In any event, even those cases cited by the Court as applying the “somewhat reduced” standard to procedural holdings still felt the need to set forth special factors justifying the overruling. United States v. Gaudin, 515 U. S. 506, 521 (1995), concluded that “the decision in question had been proved manifestly erroneous, and its underpinnings eroded, by subsequent decisions of this Court”; and Payne v. Tennessee, 501 U. S. 808, 828-830 (1991), noted that the overruled cases had been “decided by the narrowest of margins, over spirited dissents challenging [their] basic underpinnings,” had been “questioned by Members of the Court in later decisions,” and had “defied consistent application by the lower courts.”
The Court’s next excuse is that House was decided without full briefing or argument. The sole precedent it cites for the proposition that this makes a difference is Gray v. Mississippi, 481 U. S. 648, 651, n. 1 (1987). Gray, however, did not deny stare decisis effect to an opinion rendered without full briefing and argument — it accorded stare decisis effect. Id., at 666-667. What the Court relies upon is the mere dictum, rendered in the course of this opinion (and dictum in a footnote, at that), that “summary action here does not have the *260same precedential effect as does a case decided upon Ml briefing and argument.” Id., at 651, n. 1. But tbe sole authority cited for that dictum was Edelman v. Jordan, 415 U. S. 651 (1974), which declined to give stare decisis effect, not to opinions that had been issued without briefing and argument, but to judgments that had been issued without opinion — “summary affirmances” that did not “contain any substantive discussion” of the point at issue or any other point, id., at 670-671. Such judgments, affirming without comment the disposition appealed from, were common in the days when this Court had an extensive mandatory jurisdiction; they carried little more weight than denials of certio-rari. House, by contrast, was a six-page opinion with substantive discussion on the point at issue here. It reasoned: (1) “Our authority... extends only to cases In a circuit court of appeals ....’” (2) “Here the case was never fin’ the court of appeals,” because of (3) “want of a certificate of probable cause.” 324 U. S., at 44.2 And it cited as authority Ferguson v. District of Columbia, 270 U. S. 633 (1926). The new rule that the Court today announces — that our opinions rendered without Ml briefing and argument (hitherto thought to be the strongest indication of certainty in the outcome) have a diminished stare decisis effect — may well turn out to be the principal point for which the present opinion will be remembered. It can be expected to affect the treatment of many significant per curiam opinions by the lower courts, and the willingness of Justices to undertake summary disposition in the future.
*261Of course even if one accepts that the two factors the Court alludes to (procedural ruling plus absence of full briefing or argument) reduce House’s stare decisis effect, one must still acknowledge that its stare decisis effect is increased by the fact that it was a statutory holding. The Court does not contend that stare decisis is utterly inapplicable, and so it must come up with some reason for ignoring it. Its reason is that we have “disregarded” House in practice. Ante, at 252. The opinions it cites for this proposition, however, not only fail to mention House; they fail to mention the jurisdictional issue to which House pertains. And “we have repeatedly held that the existence of unaddressed jurisdictional defects has no precedential effect.” Lewis v. Casey, 518 U. S. 343, 352, n. 2 (1996) (emphasis added). Surely it constitutes “precedential effect” to reduce the stare decisis effect of one of the Court’s holdings. It is significant, moreover, that when Members of the Court have discussed House or the jurisdictional effect of a COA denial, they have agreed that jurisdiction is not available under § 1254. See Davis v. Jacobs, 454 U. S. 911, 912 (1981) (Stevens, J., respecting denial of certiorari); id., at 916-917 (Rehnquist, J., joined by Burger, C. J., and Powell, J., dissenting); Jeffries v. Barksdale, 453 U. S. 914, 915-916 (1981) (Rehnquist, J., joined by Burger, C. J., and Powell, J., dissenting). The Court’s new approach to unaddressed jurisdictional defects is perhaps the second point for which the present opinion will be remembered.
While there is scant reason for denying stare decisis effect to House, there is special reason for according it: the reliance of Congress upon an unrepudiated decision central to the procedural scheme it was creating. Section 102 of AEDPA continues a long tradition of provisions enacted by Congress that limit appellate review of petitions. In 1908, Congress required a certificate of probable cause in habeas corpus eases involving state prisoners before an appeal would lie to *262this Court, see Act of Mar. 10, 1908, ch. 76, 35 Stat. 40. In 1925, this requirement was extended to intermediate appellate proceedings, see Act of Feb. 18,1925, ch. 229, §§ 6(d), 13, 43 Stat. 940, 942. Before 1925, this Court readily concluded it had no jurisdiction over appeals brought before it in the absence of a certificate, see, e. g., Bilik v. Strassheim, 212 U. S. 551 (1908); Ex parte Patrick, 212 U. S. 555 (1908), and House interpreted the 1925 amendment to produce the same effect in the courts of appeals and, consequently, in this Court under the predecessor to § 1254(1). Quite obviously, with House on the books — neither overruled nor even cited in the later opinions that the Court claims “disregarded” it— Congress presumably anticipated that §102 of AEDPA would be interpreted in the same manner.3 In yet another striking departure from our ordinary practice, the Court qualifies the rule that statutes are deemed to adopt the extant holdings of this Court, see Keene Corp. v. United States, 508 U. S. 200, 212 (1993): They will not be deemed to adopt them, the Court says, when legal commentators “question *263the vitality” of the holdings. Ante, at 252. The confusion that will be introduced by this new approach is obvious.
At bottom, the only justification for the Court’s holding— and the only one that prompts the concurrence to overrule House — is convenience: it “permits us to carry out our normal function” of appellate review. Ante, at 251. Our “normal” function of appellate review, however, is no more and no less than what Congress says it is. U. S. Const., Art. Ill, § 2. The Court’s defiance of the scheme created by Congress in evident reliance on our precedent is a display not of “common sense,” ante, at 250, but of judicial willfulness. And a doctrine of stare decisis that is suspended when five Justices find it inconvenient (or indeed, as the concurrence suggests, even four Justices in search of a fifth) is no doctrine at all, but simply an excuse for adhering to cases we like and abandoning those we do not.
II
Since I find no jurisdiction under § 1254(1), I must address the Government’s further argument that we can issue a common-law writ of certiorari under the All Writs Act, 28 U. S. C. § 1651. The All Writs Act provides that “[t]he Supreme Court... may issue all writs necessary or appropriate in aid of [its] jurisdietio[n] and agreeable to the usages and principles of law.” As expressly noted in this Court’s Rule 20.1, issuance of a writ under § 1651 “is not a matter of right, but of discretion sparingly exercised,” and “[t]o justify the granting of any such writ, the petition must show that the writ will be in aid of the Court’s appellate jurisdiction, that exceptional circumstances warrant the exercise of the Court’s discretionary powers, and that adequate relief cannot be obtained in any other form or from any other court.”
Petitioner (who filed a petition for a writ of certiorari under § 1254(1), not under the All Writs Act, Pet. for Cert. 1) has failed to establish that he meets these requirements. To begin with, he has not shown that adequate relief is unobtainable in any form or from any other court. AEDPA dif*264fers from the gatekeeping statute at issue in House in a crucial respect: when House was decided, claimants could seek certificates of probable cause only from “the United States court by which the final decision was rendered or a judge of the circuit court of appeals,” 28 U. S. C. §466 (1940 ed.), whereas § 102 of AEDPA permits claimants to seek COA’s from a “circuit justice or judge.” Because petitioner may obtain the relief he seeks from a circuit justice, relief under the All Writs Act is not “necessary.”
Relief under the Act is also not “appropriate.” The only circumstance alleged by petitioner to justify relief is that the Eighth Circuit erroneously concluded that he failed to present a substantial constitutional question. There is nothing “exceptional” about this claim; it is in fact the same claim available to every petitioner when a COA is denied, and entertaining it would render application for this “extraordinary” writ utterly routine. Issuance of the writ is not “appropriate” for another reason as well: It would frustrate the purpose of AEDPA, which is to prevent review unless a COA is granted. ‘Where a statute specifically addresses the particular issue at hand, it is that authority, and not the All Writs Act, that is controlling. Although that Act empowers federal courts to fashion extraordinary remedies when the need arises, it does not authorize them to issue ad hoe writs whenever compliance with statutory procedures appears inconvenient or less appropriate.” Pennsylvania Bureau of Correction v. United States Marshals Service, 474 U. S. 34, 43 (1985).4
* * *
The purpose of AEDPA is not obscure. It was to eliminate the interminable delays in the execution of state and federal criminal sentences, and the shameful overloading of *265our federal criminal justice system, produced by various aspects of this Court’s habeas corpus jurisprudence. And the purpose of the specific provision of AEDPA at issue here is also not obscure: It was designed, in intelligent reliance upon a holding of this Court, to end §2255 litigation in the district court unless a court of appeals judge or the circuit justice finds reasonable basis to appeal. By giving literally unprecedented meaning to the words in two relevant statutes, and overruling the premise of Congress’s enactment, the Court adds new, Byzantine detail to a habeas corpus scheme Congress meant to streamline and simplify. I respectfully dissent.
The Court has no response to this. Its observation that a dispute over venue is not unreviewable simply because it is preliminary, ante, at 248-249, is accurate but irrelevant. The issue is not whether a venue dispute may be reviewed at all, but whether it may be reviewed in isolation from some ease of which it is a part. It may not, because a venue dispute, standing alone — like a request for a COA, standing alone — lacks the requisite qualities of a case. If the entire §2255 proceeding was not “in” the Court of Appeals, the COA request alone was not a “case” that § 1254 authorizes us to review.
The concurrence asserts that this analysis was “virtually unreasoned.” Ante, at 253 (opinion of Souter, J.). It seems to me, to the contrary, that there was virtually nothing more to be said. Not until today has anyone thought that a “ease” could consist of a disembodied request to appeal. The concurrence joins the Court in relying upon a truly eccentric argument, and then blames the House Court for not discussing this eccentricity at length.
The Court points to the fact that another provision of AEDPA, which requires court of appeals authorization before a state prisoner can file a second or successive habeas petition in district court, specifically states that the denial of the authorization “shall not be appealable and shall not be the subject of a petition ... for a writ of certiorari.” 28 U. S. C. § 2244(b)(3)(E) (1994 ed., Supp. II). This provision, the Court says, would be rendered “superfluous” if we followed House, ante, at 249. That is not so. Section 2244(b)(3) addresses whether there will be district-court consideration of a second or successive petition at all, not whether the district court’s consideration may be reviewed by an appellate court. Only the latter is covered by the holding of House. It is true enough that the reasoning of House, if carried over to the other question, would produce the same result; but Congress’s specification of that result when there is no Supreme Court holding precisely in point would more accurately be described as cautious than superfluous. Indeed, the greater relevance of § 2244(b)(3) to the question before us is this: It would be exceedingly strange to foreclose certiorari review of the denial of all federal intervention, as that provision does, while according certiorari review of the denial of appeal from the federal district court to the court of appeals.
Because petitioner has not demonstrated that issuance of the writ is “necessary” or “appropriate” under § 1651,1 need not discuss whether it fails the further requirement that it be “in aid of” our jurisdiction.