We must decide whether we should vacate, as improvidently granted, a Certificate of Appealability issued by a motions panel pursuant to the Antiterrorism and Effective Death Penalty Act.

I

Kevin Phelps was convicted of first degree murder in a California state court in 1995 and has exhausted all relevant state remedies. On May 15, 1998, he filed a habeas corpus petition in the Northern District of California one year and fifteen days after the California Supreme Court declined to review a previous denial of state habeas corpus relief. The district court denied the petition based upon its having been lodged more than a year after his state post-conviction relief process ended. See 28 U.S.C. § 2244 (d). On appeal, we affirmed in an unpublished disposition. See Phelps v. Alameda, No. 99-15495, 2000 WL 329180 (9th Cir. Mar. 29, 2000).

Two years later, Phelps filed a motion under Federal Rule of Civil Procedure 60(b)(5) with the same district court, seeking reconsideration of the earlier denial of his petition for writ of habeas corpus. He argued that Bunney v. Mitchell, 262 F.3d 973 (9th Cir.2001), among other cases, effected an intervening change in the law suggesting that his initial petition had indeed been timely. See Fed.R.Civ.P. 60(b)(5) (allowing a district court to revise an order if “a prior judgment upon which it is based has been reversed or otherwise vacated”).

The district court squarely rejected Phelps’s motion on the merits:

The court added, however, that “the Ninth Circuit has held that where a habeas corpus petitioner files a motion for relief from a final judgment under Rule 60(b), based on a subsequent change in law, the motion should be construed as a successive application for a writ of habeas corpus.” Consequently, the district court ruled, in the alternative, that it “lack[ed] subject matter jurisdiction to determine whether the change in the law would entitle petitioner to relief.”

Phelps thereupon filed an application for a Certificate of Appealability (“COA”), which the district court denied, reciting that there was no “substantial showing of the denial of a constitutional right.” See § 2258(c). Phelps timely appealed and applied for a COA from us under Federal Rule of Appellate Procedure 22(b). A motions panel granted a COA, but specifically styled the issue as “whether the district court erred in construing petitioner’s motion to reinstate his 28 U.S.C. § 2254 petition pursuant to Fed. R. Civ. P[ ]. 60(b)(5) as a successive application under 28 U.S.C. § 2244 (b).” Phelps v. Alameda, No. 02-15821 (9th Cir. Dec. 2, 2002) (order granting partial COA).

II

As a threshold matter, we must confess that we have some doubt as to our jurisdic tion over this appeal. See infra Part III. But to what extent, as a merits panel, are we bound by the operative COA, and are we compelled to decide the issue presented by it?

In federal habeas corpus proceedings, of course, the exercise of appellate jurisdiction is dependent entirely upon the issuance of a COA. 28 U.S.C. § 2253 (c); Lord v. Lambert, 347 F.3d 1091, 1094 (9th Cir.2003). The circuits, however, are split on the question of whether the COA need be examined in every case. Compare United States v. Cepero, 224 F.3d 256, 261-62 (3d Cir.2000) (en banc) (stating that “we must reject the analysis of our sister circuits” because “[t]he issuance of the certificate in the case before us is not merely an exercise of judicial gate-keeping, but rather, in the language of the [Supreme] Court, is ‘the judicial determination of a case or controversy, reviewable on appeal to the Court of Appeals’ ” (quoting Hohn v. United States, 524 U.S. 236, 246 , 118 S.Ct. 1969 , 141 L.Ed.2d 242 (1998) in turn quoting Ex parte Quirin, 317 U.S. 1, 24 , 63 S.Ct. 2 , 87 L.Ed. 3 (1942))), with Soto v. United States, 185 F.3d 48, 52 (2d Cir.1999) (“[A] certificate of appealability that does not meet the denial of a constitutional right requirement — and hence, is erroneously issued — nevertheless suffices to confer appellate jurisdiction.”), United States v. Talk, 158 F.3d 1064, 1068 (10th Cir.1998) (“[Although the absence of a certificate precludes an appeal, an erroneously-issued certificate does not deprive us of jurisdiction to hear a certified appeal.”), and Young v. United States, 124 F.3d 794, 799 (7th Cir.1997) (“The absence of a certificate of appealability precludes an appeal; should an erroneously issued certificate be treated the same as the lack of a certificate? We think not.”); cf. Peguero v. United States, 526 U.S. 23 , 119 S.Ct. 961 , 143 L.Ed.2d 18 (1999) (ruling on the merits of a claim for collateral relief even though the government pointed out in its brief that the COA may have been improperly granted).

We follow the majority view that merits panels are not required to examine allegedly defective COAs in the face of jurisdictional challenges. While “the issuance of a certificate of appealability is a prerequisite to our assertion of jurisdiction, once that certificate is issued, we have jurisdiction even if the certificate was arguably ‘improvidently granted.’ ” James v. Giles, 221 F.3d 1074, 1076 (9th Cir.2000); see also Gatlin v. Madding, 189 F.3d 882, 887 (9th Cir.1999) (holding that once a COA “has been issued without objection by this court, the procedural threshold for appellate jurisdiction has been passed and we need not revisit the validity of the certificate in order to reach the merits”).

However, we do not appear directly to have decided the related question of whether a panel has the power to address the propriety of a COA of its own accord. Our fellow circuits appear to be split on this particular question as well: Compare LaFevers v. Gibson, 182 F.3d 705, 711 (10th Cir.1999) (comparing COAs to the old Certificates of Probable Cause in holding that once “the district court has made appealable all the issues in this case by its blanket order, we must review the merits of each claim”), with Khaimov v. Crist, 297 F.3d 783, 786 (8th Cir.2002) (holding that “circumscribing, and even revoking, a certificate, especially one we have issued, is ... well within our authority”), and United States v. Marcello, 212 F.3d 1005, 1007-08 (7th Cir.2000) (holding that “the best approach is to say we have discretion to decide the case by reviewing the validity of the C[O]A”).

Section 2253(c) specifies only three things about COAs: that they must issue before an appeal may be taken, § 2253(c)(1), that they may be granted only upon “a substantial showing of the denial of a constitutional right,” § 2253(c)(2), and that they must specify the issues to be considered on appeal, § 2253(c)(3). Unfortunately, then, § 2253(c) itself provides no insight into the question of whether and to what extent we may review a previously issued COA.

Our own relevant precedent, however, provides key insight into this question. When first confronted with the reviewability of COAs in Gatlin , we took a narrow approach, declining to address the propriety of a COA upon the motion of a party because there had been no challenge to it within the appropriate 35 day window as set forth by our Circuit Rule 22-l(e). See Gatlin, 189 F.3d at 887 (“Given a ready procedure for response by the state to requests for COAs in this court, it would be counter-productive to require merits panels to reconsider every one that is issued.”).

But at the time the COA had been issued in the James case, our circuit rules had not yet provided for this 35 day opportunity to challenge a COA. See James, 221 F.3d at 1077 (“The State here argues that, because James’ motion in this court for a certificate of appealability was made prior to the adoption of 9th Cir. R. 22-l(c), the State cannot be bound by a failure to file a response.”). As a result, our merits panel proceeded to consider whether the COA properly was issued by the district court. See id. at 1077-79 (upholding issuance of the COA). If the propriety of the COA were entirely unreviewable, such action presumably would not have been appropriate, regardless of whether Circuit Rule 22-1(c) was on the books. Thus, we are satisfied that COAs are not beyond merits panel scrutiny.

This conclusion is further supported by a closely analogous principle: We do have the power to expand the scope of a COA to include additional issues, even if they previously had been deemed inappropriate for review. See, e.g., Nardi v. Stewart, 354 F.3d 1134, 1136-40 (9th Cir.2004) (expanding a COA to include a claim that both the district court and a motions panel previously had declined to certify); Valerio v. Crawford, 306 F.3d 742, 764 (9th Cir.2002) (“[A] court of appeals not only has the power to grant a COA where the district court has denied it as to all issues, but also to expand a COA to include additional issues when the district court has granted a COA as to some but not all issues.”); Hiivala v. Wood, 195 F.3d 1098, 1104 (9th Cir.1999) (“A merits panel may ... expand the issues for review to include issues that the motions panel specifically rejected.”). These cases conclusively demonstrate that the issuance of a COA is not entirely insulated from subsequent judicial scrutiny. And the fact that we have the power to grant or to expand a COA strongly implies that we have the commensurate power to vacate or to contract it.

Of course, we must be ever mindful of the “gatekeeping and efficiency functions of the certificate of appealability.” James, 221 F.3d at 1079 . In many cases, our examination of the adequacy of a COA simply does not further these goals, and the effective deployment of substantial legal resources favors turning directly to the merits. This may be particularly true either because no one has challenged the COA, see id. (declining to address the issuance of a COA “on a ground never raised by the parties”), or because the parties have already fully briefed the issues it encompasses, see Buie v. McAdory, 322 F.3d 980, 982 (7th Cir.2003) (stating that review of a COA is not often appropriate once “the appeal has been fully briefed by the time the respondent gets around to moving to vacate the certificate of appeala-bility”).

Nevertheless, the pursuit of efficiency alone does not support an absolute bar against examining the validity of a COA. In many instances, few legal resources may have been spent by the time a party challenges a COA. So if “briefing has not yet begun but the certificate has identified a constitutional issue of dubious substan-tiality,” it may make a good deal of sense to consider a challenge to a COA. Buie, 322 F.3d at 982 ; cf. 9th Cir. R. 22-1(c) & (d) (providing a mechanism by which a party may challenge whether a COA should issue or be expanded).

Moreover, there may be competing concerns involved, and in exceptional circumstances the vacatur of a COA may be appropriate regardless of the investment of timé and energy into the case. For example, the issuance of a COA may have been so far off the mark that the certificate is simply invalid on its face. If we had no power to vacate COAs, we would be unable adequately to participate in the proper administration of § 2253(c). The decision to grant a COA would “effectively be unreviewable on appeal,” a highly disfavored result. Batzel v. Smith, 333 F.3d 1018, 1025 (9th Cir.2003) (asserting jurisdiction in an interlocutory appeal in part because to do otherwise would insulate the lower court’s ruling); see also Ramunno v. United States, 264 F.3d 723, 725 (7th Cir.2001) (“Vacating a certificate of appealability is an unusual step ... but the possibility of review is essential if the statutory limits are to be implemented. Otherwise district judges have the authority to issue certificates of appealability for any reason at all, and as open-ended as they please.”).

Based on our review of the relevant precedent, we are satisfied that although a merits panel generally need not examine the propriety of a COA, it nevertheless retains the power to do so.

III

And now to our jurisdictional dilemma in this case: Irrespective of the “successive petition” alternative holding, the district court simply denied Phelps’s Rule 60(b) motion on the merits, a ruling as to which no COA has been granted. Thus, any decision by us would offer no relief. If the district court’s successive petition ruling were indeed correct, the dismissal of Phelps’s motion would stand. If the successive application ruling were erroneous, the result would remain the same: Phelps’s motion would still be unsuccessful because the district court has already denied it on the merits. In other words, no matter how we rule on appeal, Phelps’s Rule 60(b) motion fails.

A

Alternative holdings may be valuable tools of judicial administration, but they can present potential advisory-opinion problems. See, e.g., Karsten v. Kaiser Found. Health Plan, 36 F.3d 8, 11-12 (4th Cir.1994) (noting that alternative holdings often “provide courts, particularly appellate courts reviewing alternative holdings below, with the tempting opportunity to stray into the practice of advisory opinion-making, solving questions that do not actually require answering in order to resolve the matters before them”). Normally, such forays onto what might be described as thin jurisprudential ice are acceptable consequences of our continuing pursuit of justice and finality. Cf., e.g., Harris v. Reed, 489 U.S. 255 , 264 n. 10, 109 S.Ct. 1038 , 103 L.Ed.2d 308 (1989) (noting that by use of alternative holdings, “a state court may reach a federal question without sacrificing its interests in finality, federalism, and comity”).

But habeas corpus petition cases are unique: Under AEDPA, we simply “lack jurisdiction to resolve the merits of any claim for which a COA is not granted.” Beaty v. Stewart, 303 F.3d 975, 984 (9th Cir.2002). Consequently, we may be presented with a case, such as this one, where resolution of any issue on appeal may not “affect the matter in issue in the case before” us. Mills v. Green, 159 U.S. 651, 653 , 16 S.Ct. 132 , 40 L.Ed. 293 (1895).

In this respect, the limited jurisdiction preserved by the grant of a COA is not unlike the Supreme Court’s authority over state-court decisions. Because the high Court does not exercise jurisdiction over purely state-law issues, an “adequate and independent” state-law ground for decision essentially makes any federal issues superfluous, and the Court will not consider them. See Michigan v. Long, 463 U.S. 1032, 1041 , 103 S.Ct. 3469 , 77 L.Ed.2d 1201 (1983) (“If the state court decision indi cates clearly and expressly that it is alternatively based on bona fide separate, adequate, and independent grounds, we, of course, will not undertake to review the decision.”). This approach “avoid[s] the danger of ... rendering advisory opinions.” Id.; see also Local No. 8-6, Oil, Chem. & Atomic Workers Int’l Union v. Missouri, 361 U.S. 363, 367 , 80 S.Ct. 391 , 4 L.Ed.2d 373 (1960) (“[T]he duty of this Court is to decide actual controversies by a judgment which can be carried into effect, and not to give opinions upon moot questions or abstract propositions, or to declare principles or rules of law which cannot affect the matter in issue in the case before it.” (internal quotation omitted)).

B

Here, we are confronted with an “adequate and independent” federal ground for decision outside our jurisdic: tional purview. As a result, consideration of the issue specified by the COA in this case is inappropriate. See Kaminski v. United States, 339 F.3d 84 , 85 n. 1 (2d Cir.2003) (asserting, in habeas corpus context, that “ordinarily, unless a certificate [of appealability] encompasses all of the grounds for a court’s ruling on an issue, an appeal that challenges only some of the district court’s grounds will be moot” ).

Perhaps this problem might have been avoided had Phelps sought to expand the COA to include all dispositive holdings. But “[i]f a party wishes to expand the scope of a partial COA, he must follow the procedure set forth in Circuit Rule 22 1(d), which requires him to seek and obtain from the appellate court broader certification.” United States v. Christakis, 238 F.3d 1164 , 1168 n. 5 (9th Cir.2001). Not only has Phelps failed to seek certification of the district court’s alternative holding, he has affirmatively argued against it in both his opening and reply briefs, asserting that “[a]ny other argument, such as respondent’s claim that petitioner is not entitled to Rule 60(b) relief, is not before this court.”

In this case, if we were to entertain the only question presented to us, we would be required to render a ruling without “any effectual relief whatever,” Mills, 159 U.S. at 653 , 16 S.Ct. 132 , as we are without jurisdiction to affect the denial of Phelps’s Rule 60(b) motion.

IV

For the foregoing reasons, we vacate the COA as improvidently granted. Because we therefore lack subject-matter jurisdiction, the appeal is

DISMISSED.

. Because the state court's action was an order rather than an opinion, it became final upon filing. See Cal. Rules of Court 24(a) since amended, with corresponding provision now appearing in Rule 29.4(b)(2) ("An order of the Supreme Court denying a petition for review of a decision of a Court of Appeal becomes final when it is filed.”). . We use the term "merits” somewhat loosely here, for, of course, a decision to dismiss a Rule 60(b) motion means that the district court will not reconsider the underlying merits of the case. Instead, "merits” in this context relates to the court’s consideration of the substance of Phelps's claim that a Rule 60(b) motion should be granted. . Although somewhat ambiguous, we must interpret the discussion of subject-matter jurisdiction as an alternative holding for at least two reasons. First, the district court’s subsequent denial of a Certificate of Appealability nowhere mentions the jurisdictional issue, and states only that the earlier decision was "based on settled Ninth Circuit law that a change in the applicable law after a judgment has become final in all respects is not a sufficient basis for vacating a judgment under Rule 60(b)(5)” (internal quotation omitted). Second, given the district court's simultaneous denial of Phelps's Rule 60(b) motion on the merits, we must seek to avoid the constitutional tension that a contrary interpretation likely would engender. See Christopher v. Harbury, 536 U.S. 403, 417 , 122 S.Ct. 2179 , 153 L.Ed.2d 413 (2002) (highlighting "the obligation of the Judicial Branch to avoid deciding constitutional issues needlessly”); Willy v. Coastal Corp., 503 U.S. 131, 138 , 112 S.Ct. 1076 , 117 L.Ed.2d 280 (1992) (noting the constitutional "issue of a district court adjudicating the merits of a 'case or controversy’ over which it lacks jurisdiction”); Firestone Tire & Rubber Co. v. Risjord, 449 U.S. 368, 379 , 101 S.Ct. 669 , 66 L.Ed.2d 571 (1981) (holding that "[a] court lacks discretion to consider the merits of a case over which it is without jurisdiction”). We recognize more generally that an adverse jurisdictional determination stands at apparent odds with the rendering of an alternative holding on the merits of the same case. Nevertheless, we have held that "alternative holdings are a common practice that prevents the overconsumption of adjudicative resources” and "do not divest the adjudicator of jurisdiction merely because they are inconsistent.” Container Stevedoring Co. v. Dir., Office of Workers Comp. Programs, 935 F.2d 1544 , 1549 n. 5 (9th Cir.1991). . The Third Circuit also has indicated that merits panels have such authority, given that they are required to review COAs. See Cepero, 224 F.3d at 261-62 . Second Circuit jurisprudence is somewhat less clear, although relevant precedent suggests that a merits panel may at least ignore an improper COA in certain circumstances. See, e.g., Rhagi v. Artuz, 309 F.3d 103, 105 (2d Cir.2002) (dismissing a case because "no COA has been granted that would permit appellate review;” although "the District Court actually did issue a COA,” it was "deficient”). But see Valverde v. Stinson, 224 F.3d 129, 136 (2d Cir.2000) (understanding Soto as "holding that merits panel may not revisit motion panel’s decision to grant a certificate of appealability”). . It would be difficult to understand Circuit Rule 22-l(c) itself as an independent and sufficient mechanism insulating COAs from judicial review. It merely sets forth a procedure by which parties can challenge the issuance of a COA. It does not thereby act of its own accord entirely to prevent a merits panel from examining the issue if necessaiy. A party's failure to comply with Rule 22-1 (c) is more properly understood as a waiver of any challenge to a COA. And when a party waives issues on appeal, we generally do not impose an absolute barrier against their consideration. See, e.g., Koerner v. Grigas, 328 F.3d 1039, 1048-49 (9th Cir.2003) (listing exceptions to the normal rule that parties waive issues not specifically and distinctly argued in their opening briefs). . Whether originally issued by a district court or a motions panel, “rulings on jurisdictional issues do not bind a merits panel” of this Court. Hiivala, 195 F.3d at 1104 . Therefore, the "doctrine of law of the case’ does not preclude” our power to review the issuance of a COA. Id. . We note that there should be no important future consequence of the lower court’s ruling that this was a successive petition. See Hill v. Alaska, 297 F.3d 895, 898 (9th Cir.2002) ("That a prisoner has previously filed a federal habeas petition does not necessarily render a subsequent petition 'second or successive.’ "); Calderon v. U.S. Dist. Court, 163 F.3d 530 , 538 (9th Cir.1998) (en banc) abrogated on other grounds by Woodford v. Garceau, 538 U.S. 202 , 123 S.Ct. 1398 , 155 L.Ed.2d 363 (2003) ("We conclude that even after the AEDPA, the rule is as it has always been: Res judicata does not apply to habeas cases.”). . The prohibition against advisory opinions is often couched in constitutional terms, but when there are legal questions directly involving the litigants before a court — even though resolution of an issue may have no ultimate effect upon them—a disposition squarely may not implicate the Constitution. See, e.g., Long, 463 U.S. at 1041 n. 6, 103 S.Ct. 3469 (holding that even when there is an adequate and independent state ground for decision, "[t]here may be certain circumstances in which clarification is necessary or desirable, and we will not be foreclosed from taking the appropriate action”); Evan Tsen Lee, Deconstitutionalizing Justiciability: The Example of Mootness, 105 Harv.L.Rev. 603, 644-45 (1992) (arguing that the bar against "Supreme Court review of any state judgment for which there is or may be an adequate and independent state ground of decision” is "a function of judicial discretion”). But see Powell v. McCormack, 395 U.S. 486 , 496 n. 7, 89 S.Ct. 1944 , 23 L.Ed.2d 491 (1969) ("The rule that this Court lacks jurisdiction to consider the merits of a moot case is a branch of the constitutional command that the judicial power extends only to cases or controversies.”). In any case, we remain bound by a longstanding line of Supreme Court precedent refusing to issue an opinion that does not affect the outcome of the case at bar. See, e.g., Mills, 159 U.S. at 653 , 16 S.Ct. 132 . . Normally, a case is rendered moot "when, by virtue of an intervening event, a court of appeals cannot grant any effectual relief whatever.” Calderon v. Moore, 518 U.S. 149, 150 , 116 S.Ct. 2066 , 135 L.Ed.2d 453 (1996) (internal citation omitted). But in cases such as this, it is the imprudent decision to grant the COA that gives rise to the problem, not some extrajudicial event. Thus, “mootness” may not be the best term, and the issue is perhaps more appropriately described simply as a "jurisdictional” matter. Cf. Coleman v. Thompson, 501 U.S. 722, 729 , 111 S.Ct. 2546 , 115 L.Ed.2d 640 (1991) (stating "that the independent and adequate state ground doctrine is jurisdictional”).