concur, dissenting from the denial of rehearing in banc.
“A court may review a final order of removal only if ... the alien has exhausted all administrative remedies available to the alien as of right .... ” 8 U.S.C. § 1252(d)(1). Thus Congress foreclosed our review of challenges that an alien failed to present at the agency’s factfind-ing and appellate levels. The majority of a divided panel says that we can reaeh unex-hausted issues nevertheless — subject only to our discretion and to the spongy test of “manifest injustice.” The panel majority seems to think that Congress has shut the door with the intention that we should climb in the window.
Our law on this subject has been well settled to the contrary. As the panel majority concedes, “our court has consistently applied an issue exhaustion requirement to petitions for review from the BIA.” Zhong v. U.S. Dep’t of Justice, 480 F.3d 104, 119 (2d Cir.2007). Judge Kearse’s dissent persuasively attacks the deviation of the panel majority and the analysis they use to get *135where they go. I cannot improve upon her dissent, and there is no point in my repeating it. This dissent is taken from the denial of in banc review because I do not know what makes the panel majority believe that our precedent is insufficient to decide this case, or that our precedent could be set aside by one panel.
I
Our 2004 opinion in Foster v. INS held that § 1252(d)(1) precludes our review of claims and of issues that were not exhausted at the BIA: “[CJourts are not free to dispense with” the requirement that a petitioner “raise issues to the BIA in order to preserve them for judicial review.” 376 F.3d 75, 77-78 (2d Cir.2004) (emphasis in original) (internal quotation marks omitted). “We held in ... Foster that a petitioner must have raised an issue below to present it on appeal.” Gill v. INS, 420 F.3d 82, 86 (2d Cir.2005) (emphasis in original). Since Foster came down, every active judge of this Court has applied its holding — usually by summary order in recognition of the settled nature of the proposition.1 Ten other courts of appeals have addressed the issue — and they all reached the same conclusion we did.2 Judge Calabresi notes that two of the circuits reached this conclusion only because “they were bound by earlier precedents in their circuits and could do nothing else.” Concurring Op. at 129. I do not think that adherence to precedent is a value so easily dismissed; in any event, those panels could have done something else: seek review in banc.
Choosing instead to depart from our well-settled precedent, the panel majority now says that we do have the discretion to review an issue presented to us on appeal that was not presented to the BIA, and moreover that we must review such an issue if necessary to prevent a “manifest injustice” — thus reserving a power to be arbitrary and captious.
Although the statute does not speak in so many words of exhausting “issues,” Foster held that it makes no sense to read the statute any other way: “Accepting the blanket statement that removal is improper as sufficient to exhaust all claims would eviscerate th[e] [exhaustion] rule since any alien could claim (as nearly all do) he was not removable and then apply for judicial review on the theory he had exhausted his remedies.” 376 F.3d at 78. This holding has been consistently applied in our opinions. See, e.g., Iouri v. Ashcroft, 464 F.3d 172, 177 (2d Cir.2006) (“[T]his particular argument was not raised before the BIA and Petitioners therefore faded to exhaust their administrative remedies.” (emphasis added)).
*136II
Panels of this Court are bound by our precedents unless and until those precedents aré overruled by the Supreme Court or by this Court sitting in banc. See Nicholas v. Goord, 430 F.3d 652, 659 (2d Cir.2005). Nothing the Supreme Court has said could have justified the departure from Circuit precedent that the panel majority has made.
The Supreme Court has spoken plainly as to how the inferior courts should understand statutory exhaustion requirements. When Congress says (as it did in § 1252(d)(1)) that a petitioner must have “exhausted all administrative remedies available as of right,” the petitioner must do so in accordance with that agency’s procedures. See Jones v. Bock, — U.S. —, 127 S.Ct. 910, 923, 166 L.Ed.2d 798 (2007) (“[I]t is the prison’s requirements ... that define the boundaries of proper exhaustion.”). The Supreme Court has thus emphasized that the existence of an issue exhaustion requirement does not hinge solely on statutory wording: “Proper exhaustion demands compliance with an agency’s deadlines and other critical procedural rules because no adjudicative system can function effectively without imposing some orderly structure on the course of its proceedings.” Woodford v. Ngo, — U.S. —, 126 S.Ct. 2378, 2386, 165 L.Ed.2d 368 (2006).
Thus in Sims v. Apfel the Court ascertained the need for issue exhaustion by examining the agency’s regulations and its internal appellate process. 530 U.S. 103, 106-12, 120 S.Ct. 2080, 147 L.Ed.2d 80 (2000).3 Sims ruled that appeals to the Social Security Appeals Council need not specify grounds for reversal; but the analysis used in Sims compels the opposite conclusion in the immigration context. The Court concluded that social security regulations did not require issue exhaustion, but observed that “it is common for an agency’s regulations to require issue exhaustion in administrative appeals,” id. at 108, 120 S.Ct. 2080, and the example it cited of a regulation that does require as much — and therefore would necessitate issue exhaustion — closely tracks the words of its immigration analog. Compare 20 C.F.R. § 802.211(a) (petitions for review by the Benefits Review Board must “list the specific issues to be considered on appeal”), with 8 C.F.R. § 1003.3(b) (“The party taking the appeal must identify the reasons for the appeal ... [including] the findings of fact, the conclusions of law, or both, that are being challenged.”).
In considering the agency’s appellate process, the Sims Court examined the agency’s directions to those seeking to file an appeal within the agency, and inferred that the “Council does not depend much, if at all, on claimants to identify issues for review.” 530 U.S. at 111-12, 120 S.Ct. 2080 (citing 20 C.F.R. § 422.205(a) and Form HA-520). By contrast, the Executive Office for Immigration Review expressly instructs aliens to “give specific details why you disagree with the Immi*137gration Judge’s decision” and to “specify the finding(s) of fact, the conclusion(s) of law, or both, that you are challenging.” Form EOIR-26, available at http://www. usdoj.gov/eoir/; see also 8 C.F.R. § 1003.38.
Ill
Having decided to bypass our precedent, the panel majority makes a pretense of laying down criteria for the circumstances (purportedly limited) in which we might excuse or ignore a petitioner’s failure to exhaust issues before the BIA.
First, the panel majority would consider whether the government objected on appeal to the petitioner’s failure to exhaust. Zhong, 480 F.3d at 107 n. 1, 125. (This is raised as a purported ground of distinction between the present case and Foster). Even were this a legitimate consideration, no waiver should be inferred from the lack of objection by the government where, as here, the government had no reason to assume that we would deviate from our well-settled precedent. If waiver can be inferred here, it can be inferred anywhere. And future panels may simply deem an objection ‘insufficiently specific’ to render the waiver issue — which the panel majority regards as essential to considering the unexhausted issues in the present case— meaningless.
In any event, § 1252(d)(1) tells us what the “court may review,” not what the parties must do; so its mandate should not be reduced to the status of an affirmative defense.4 The statutory context demonstrates why Congress chose to express § 1252(d)(1) as a limitation on our power. Courts of appeals are empowered only to review final orders of removal, not to re-adjudicate the underlying application for relief.5 Because we cannot review something that the BIA had no occasion or duty to decide, we should not be freed from the constraint of § 1252(d)(1) by the absence of a government objection or by the government’s undertaking instead to demonstrate that an unpreserved issue is merit-less.
The panel majority would next consider whether the BIA affirmed summarily, and draws the senseless inference that a summary affirmance adopts all of the IJ’s grounds for denying relief even if the petitioner challenged fewer than all of them. Zhong, 480 F.3d at 125. But why presume that the BIA considers and decides issues that were not argued to it? And why assume that the BIA conducts a broader review before issuing a summary disposition than it does when it issues an opinion? See United States v. Gonzalez-Roque, 301 F.3d 39, 47 (2d Cir.2002) (“Although the BIA has access to the entire record, it is not obligated to search it for possibly meritorious appellate issues.”). One suspects *138that this consideration is raised by the panel majority chiefly as a way to embarrass the BIA for adopting streamlined practices. “[I]t is not the role of the federal courts to dictate the internal operating rules of the BIA.” Kambolli v. Gonzales, 449 F.3d 454, 465 (2d Cir.2006). As it happens, our Court follows a similar rule. See 2d Cir. R. § 0.23.
Last, the panel majority would consider whether “additional factfinding is necessary,” Zhong, 480 F.3d at 107 n. 1, 125. This is just another way of saying that a panel may consider on appeal any issue it sees in the record without regard to whether it was raised to the BIA. In any event, under 8 U.S.C. § 1252(a)(1), “the court may not order the taking of additional evidence.”
The considerations proffered by the panel majority amount to no effective limitation on the willingness of a panel to disregard issue exhaustion at will.6 And so, in an effort to create the illusion of restraint, the panel majority holds that the only eases in which a panel must ignore § 1252(d)(1) are those in which doing so would “prevent manifest injustice.” Zhong, 480 F.3d at 107 n. 1.
The panel majority’s footnote discussing “manifest injustice” is pure dicta because no injustice (manifest or not) has been identified in this case. In any event, I share Judge Kearse’s skepticism concerning “whether this Court actually has the power to remedy even a manifest injustice when we lack jurisdiction.” Zhong, 480 F.3d at 138 (Kearse, J., dissenting). The idea itself was improvised by a panel of this court in Marrero Pichardo v. Ashcroft, an appeal from the denial of habeas relief sought by a felon who failed altogether to appeal his order of deportation to the BIA. 374 F.3d 46, 52-54 (2d Cir.2004). Recent rulings had called into question whether Marrero Pichardo could be deemed an aggravated felon on the basis of his DUI convictions — of which there were eleven (hence the manifest injustice). As I have observed elsewhere, Marrero Pichardo can tenably stand only for two propositions already laid down by the Supreme Court: [i] “courts may not be bound by congressional limitations on jurisdiction that raise constitutional problems”; and [ii] exhaustion requirements “do not apply if there is no possibility of relief from the administrative agency, in which event the administrative officers would presumably have no authority to act on the subject of the complaint, leaving the [complainant] with nothing to exhaust.” Gill, 420 F.3d at 95-97 (Jacobs, J., dissenting) (internal citations omitted). Marrero Pichardo stands for nothing more, and certainly does not authorize us (if any panel opinion could) to reach unexhausted issues on an appeal from the BIA, or to assume that injustices that are truly “manifest” will have been overlooked by the petitioners themselves.
IV
The panel majority opinion makes its case for why § 1252(d)(1) leaves issue ex*139haustion to discretion; it makes no case for why any member of this Court is bound to follow it.
When an issue is settled (as this one has been), panels that follow precedent have no occasion to revisit it-they simply apply the rule by summary order. So judges with a mind to undo a precedent need only await such a case with patience and select themselves to decide the issue as if it had not already been settled. Such deviation from precedent is something that a disciplined court should consider in banc in order to preserve the coherence and consistency of its jurisprudence. By rejecting in banc review, we decline to consider together an issue of “exceptional importance,” one which no one can dispute is “necessary to secure or maintain uniformity” in our Circuit. Fed. R.App. P. 35(a).
Sometimes, in banc proceedings can be obviated by the circulation of a draft opinion to the active members of the Court; litigants and the public are then advised by footnote that no objection was interposed. That was not done in this case.7 This circumvention can be effective only because our in banc practice is so rusty and cumbersome that its desuetude will allow a single panel to skate past full court review.
By the same token, however, any other panel may — with equal authority and equal occasion and equal legitimacy — overrule the panel majority’s holding. This prospect is institutionally dangerous, and risks debasing the currency of our rulings and opinions.
. See, e.g., Damko v. INS, 178 Fed.Appx. 85 (2d Cir.2006); Qin Di Chen v. BCIS, 190 Fed.Appx. 101 (2d Cir.2006); Islamovic v. Gonzales, 192 Fed.Appx. 47 (2d Cir.2006); Meiying Lin v. U.S. Dep’t of Justice, 188 Fed.Appx. 28 (2d Cir.2006); Xian Xian Huang v. Gonzales, 187 Fed.Appx. 107 (2d Cir.2006); Yong Gui Liu v. Gonzales, 185 Fed.Appx. 41 (2d Cir.2006); Yun Yu Zheng v. Gonzales, 179 Fed.Appx. 752 (2d Cir.2006); Xharo v. Gonzales, 195 Fed.Appx. 28 (2d Cir.2006); Yan Fang Wang v. U.S. Dep’t of Justice, 161 Fed.Appx. 130 (2d Cir.2005).
. See Makhoul v. Ashcroft, 387 F.3d 75, 80 (1st Cir.2004); Xin Jie Xie v. Ashcroft, 359 F.3d 239, 245 n. 8 (3d Cir.2004); Asika v. Ashcroft, 362 F.3d 264, 267 n. 3 (4th Cir.2004); Kuang-Te Wang v. Ashcroft, 260 F.3d 448, 452-453 (5th Cir.2001); Ramani v. Ashcroft, 378 F.3d 554, 558-60 (6th Cir.2004); Pjetri v. Gonzales, 468 F.3d 478, 481 (7th Cir.2006); Alyas v. Gonzales, 419 F.3d 756, 762 (8th Cir.2005); Morales-Alegria v. Gonzales, 449 F.3d 1051, 1058-59 (9th Cir.2006); Rivera-Zurita v. INS, 946 F.2d 118, 120 & n. 2 (10th Cir.1991); Fernandez-Bernal v. Attorney General, 257 F.3d 1304, 1317 n. 13 (11th Cir.2001).
. The Sims Court consulted the regulations to determine the propriety of an exhaustion rule because the statute at issue in that case made no reference to exhausting administrative remedies, see 42 U.S.C. § 405(g) (requiring only a "final decision”). While Sims is concerned with judicial imposition of an issue exhaustion requirement, the Court’s analysis of agency processes should inform our determination of the scope of a statutory exhaustion requirement such as § 1252(d)(1). I agree with Judge Kearse that Foster reached the right answer in making this determination. See Zhong, 480 F.3d at 137 (Kearse, dissenting) ("Given the present context, to wit, issues presented in a petition for review, I regard the phrase 'issues not raised in the course of exhausting all administrative remedies’ (emphases added) as an oxymoron.”).
. Cf. Eberhart v. United States, 546 U.S. 12, 15-16, 126 S.Ct. 403, 163 L.Ed.2d 14 (2005) (holding that a violation of Federal Rule of Criminal Procedure 33's deadline for the filing of a motion for a new trial, which is prefaced with "Any motion ... must be filed,” must be timely objected to); Johnson v. Testman, 380 F.3d 691, 695 (2d Cir.2004) (holding that the exhaustion requirement of the Prison Litigation Reform Act, 42 U.S.C. § 1997e(a), which is prefaced with "No action shall be brought,” is waivable).
. Compare 28 U.S.C. § 2349(a) (limiting our jurisdiction over final orders of removal to "determining the validity of, and enjoining, setting aside, or suspending, in whole or in part, the order of the agency”), and 8 U.S.C. § 1252(a)(1) (incorporating by reference the jurisdictional limitations of the Administrative Orders Review Act, 28 U.S.C. § 2341 et seq.), with 42 U.S.C. § 405(g) (permitting a federal court to issue "a judgment affirming, modifying, or reversing the decision of the Commissioner of Social Security ... without remanding the cause for a rehearing”).
. Judge Calabresi offers comfort in the form of 26 cases decided after Zhong came down in which panels did not grasp the opportunity to decide a question never presented to the BIA. Concurring Op. at 133-34. This score is offered as a response to a floodgate argument I do not make. The institutional danger is not that many panels will take the opportunity to decide many such issues; it is that a single panel will be able to designate itself to decide a particular issue, and reach it at will after dispensing with the essentially meaningless criteria the Zhong majority formulated as a supposed restraint on the abuse of power and discretion. In any event, the score cited reflects only what has happened thus far — while this in banc initiative has been pending. Panels may be less inclined to exercise judicial restraint in the future. Stay tuned.
. Judge Calabresi advises that two of the three judges who signed the per curiam in Foster are willing to concede that the most significant ruling in it may be dicta. Concurring Op. at 128. I do not think that the informal approbation of two possible authors is a substitute for in banc review of a case whose principle has been consistently cited as authority by this Court, relied upon by litigants, and adopted by ten other courts of appeals.