dissenting.
The Court may be correct when it says that Congress cannot constitutionally exercise removal authority over an official vested with the budget-reduction powers that § 251 of the Balanced Budget and Emergency Deficit Control Act of 1985 *777gives to the Comptrollér General. This, however, is not because “the removal powers over the Comptroller General’s office dictate that he will be subservient to Congress,” ante, at 730; I agree with Justice White that any such claim is unrealistic. Furthermore, I think it is clear under Humphrey’s Executor v. United States, 295 U. S. 602 (1935), that “executive” powers of the kind delegated to the Comptroller General under the Deficit Control Act need not be exercised by an officer who serves at the President’s pleasure; Congress certainly could prescribe the standards and procedures for removing the Comptroller General. But it seems to me that an attempt by Congress to participate directly in the removal of an executive officer — other than through the constitutionally prescribed procedure of impeachment — might well violate the principle of separation of powers by assuming for Congress part of the President’s constitutional responsibility to carry out the laws.
In my view, however, that important and difficult question need not be decided in this litigation, because no matter how it is resolved the plaintiffs, now appellees, are not entitled to the relief they have requested. Appellees have not sought invalidation of the 1921 provision that authorizes Congress to remove the Comptroller General by joint resolution; indeed, it is far from clear they would have standing to request such a judgment. The only relief sought in this case is nullification of the automatic budget-reduction provisions of the Deficit Control Act, and that relief should not be awarded even if the Court is correct that those provisions are constitutionally incompatible with Congress’ authority to remove the Comptroller General by joint resolution. Any incompatibility, I feel, should be cured by refusing to allow congressional removal — if it ever is attempted — and not by striking down the central provisions of the Deficit Control Act. However wise or foolish it may be, that statute unquestionably ranks among the most important federal enactments of the past several *778decades. I cannot see the sense of invalidating legislation of this magnitude in order to preserve a cumbersome, 65-year-old removal power that has never been exercised and appears to have been all but forgotten until this litigation.1
*779I
The District Court believed it had no choice in this matter. Once it concluded that the Comptroller General’s functions under the Deficit Control Act were constitutionally incompatible with the 1921 removal provision, the District Court considered itself bound as a matter of orderly judicial procedure to set aside the statute challenged by the plaintiffs. See Synar v. United States, 626 F. Supp. 1374, 1393 (DC 1986). The majority today does not take this view, and I believe it is untenable.
Under the District Court’s approach, everything depends on who first files suit. Because Representative Synar and *780the plaintiffs who later joined him in this case objected to budget cuts made pursuant to the Deficit Control Act, the District Court struck down that statute, while retaining the 1921 removal provision. But if the Comptroller General had filed suit 15 minutes before the Congressman did, seeking a declaratory judgment that the 1921 removal power could not constitutionally be exercised in light of the duties delegated to the Comptroller General in 1985, the removal provision presumably would have been invalidated, and the Deficit Control Act would have survived intact. Momentous issues of public law should not be decided in so arbitrary a fashion. In my view, the only sensible way to choose between two conjunctively unconstitutional statutory provisions is to determine which provision can be invalidated with the least disruption of congressional objectives.
The District Court apparently thought differently in large part because it believed this Court had never undertaken such analysis in the past; instead, according to the District Court, this Court has “set aside that statute which either allegedly prohibits or allegedly authorizes the injury-in-fact that confers standing upon the plaintiff.” 626 F. Supp., at 1393. But none of the four cases the District Court cited for this proposition discussed the problem of choice of remedy, and in none of them could a strong argument have been made that invalidating the other of the inconsistent statutory provisions would have interfered less substantially with legislative goals or have been less disruptive of governmental operations.2
*781More importantly, the District Court ignored what appears to be the only separation-of-powers case in which this Court did expressly consider the question as to which of two incompatible statutes to invalidate: Glidden Co. v. Zdanok, 370 U. S. 530 (1962). The petitioners in that case had received unfavorable rulings from judges assigned to temporary duty in the District Court or Court of Appeals from the Court of Claims or the Court of Customs and Patent Appeals; they argued that those rulings should be set aside because the judges from the specialized courts did not enjoy the tenure and compensation guaranteed by Article III of the Constitution. Before the assignments, Congress had pronounced the Court of Claims and the Court of Customs and Patent Appeals to be Article III courts, implying that judges on those courts were entitled to Article III benefits. Older statutes, however, gave both courts authority to issue advisory opinions, an authority incompatible with Article III status. Glidden held that the Court of Claims and the Court of Customs and Patent Appeals were indeed Article III tribunals. With respect to the advisory-opinion jurisdiction, Justice Harlan’s opinion for the plurality noted: “The overwhelming majority of the Court of Claims’ business is composed of cases and controversies.” 370 U. S., at 583. Since *782“it would be . . . perverse to make the status of these courts turn upon so minuscule a portion of their purported functions,” Justice Harlan reasoned that, “if necessary, the particular offensive jurisdiction, and not the courts, would fall.” Ibid. Justice Clark’s opinion concurring in the result for himself and the Chief Justice similarly concluded that the “minuscule” advisory-opinion jurisdiction of the courts in question would have to bow to the Article III status clearly proclaimed by Congress, and not vice versa. Id., at 587-589.
The Court thus recognized in Glidden that it makes no sense to resolve the constitutional incompatibility between two statutory provisions simply by striking down whichever provision happens to be challenged first. A similar recognition has underlain the Court’s approach in equal protection cases concerning statutes that create unconstitutionally circumscribed groups of beneficiaries. The Court has noted repeatedly that such a defect may be remedied in either of two ways: the statute may be nullified, or its benefits may be extended to the excluded class. See, e. g., Heckler v. Mathews, 465 U. S. 728, 738 (1984); Califano v. Westcott, 443 U. S. 76, 89 (1979). Although extension is generally the preferred alternative, we have instructed lower courts choosing between the two remedies to “ ‘measure the intensity of [legislative] commitment to the residual policy and consider the degree of potential disruption of the statutory scheme that would occur by extension as opposed to abrogation.’ ” Heckler v. Mathews, supra, at 739, n. 5, quoting Welsh v. United States, 398 U. S. 333, 365 (1970) (Harlan, J., concurring in result). Calculations of this kind are obviously more complicated when a court is faced with two different statutes, enacted decades apart, but Glidden indicates that even then the task is judicially manageable. No matter how difficult it is to determine which remedy would less obstruct congressional objectives, surely we should make that determination as best we can instead of leaving the selection to the litigants.
*783h-H H-t
Assuming that the Comptroller General s functions under § 251 of the Deficit Control Act cannot be exercised by an official removable by joint resolution of Congress, we must determine whether legislative goals would be frustrated more by striking down § 251 or by invalidating the 1921 removal provision. That question is not answered by the “fallback” provisions of the 1985 Act, which take effect “[i]n the event that any of the reporting procedures described in section 251 [of the Act] are invalidated.” § 274(f)(1), 99 Stat. 1100. The question is whether the reporting procedures should be invalidated in the first place. The fallback provisions simply make clear that Congress would prefer a watered-down version of the Deficit Control Act to none at all; they provide no evidence that Congress would rather settle for the watered-down version than surrender its statutory authority to remove the Comptroller General. The legislative history of the Deficit Control Act contains no mention of the 1921 statute, and both Houses of Congress have argued in this Court that, if necessary, the removal provision should be invalidated rather than §251. See Brief for Appellant United States Senate 31-43; Brief for Appellants Speaker and Bipartisan Leadership Group of United States House of Representatives 49; accord, Brief for Appellant Comptroller General 33-47. To the extent that the absence of express fallback provisions in the 1921 statute signifies anything, it appears to signify only that, if the removal provision were invalidated, Congress preferred simply that the remainder of the statute should remain in effect without alteration.3
*784In the absence of express statutory direction, I think it is plain that, as both Houses urge, invalidating the Comptroller General’s functions under the Deficit Control Act would frustrate congressional objectives far more seriously than would refusing to allow Congress to exercise its removal authority under the 1921 law. The majority suggests that the removal authority plays an important role in furthering Congress’ desire to keep the Comptroller General under its control. But as Justice White demonstrates, see ante, at 770-773, the removal provision serves feebly for such purposes, especially in comparison to other, more effective means of supervision at Congress’ disposal. Unless Congress institutes impeachment proceedings — a course all agree the Constitution would permit — the 1921 law authorizes Congress to remove the Comptroller General only for specified cause, only after a hearing, and only by passing the procedural equivalent of a new public law. Congress has never attempted to use this cumbersome procedure, and the Comptroller General has shown few signs of subservience.4 If Congress in 1921 *785wished to make the Comptroller General its lackey, it did a remarkably poor job.
Indeed, there is little evidence that Congress as a whole was very concerned in 1921 — much less in 1985 or during the intervening decades — with its own ability to control the Comptroller General. The Committee Reports on the 1921 Act and its predecessor bills strongly suggest that what was critical to the legislators was not the Comptroller General’s subservience to Congress, but rather his independence from the President. See, e. g., H. R. Rep. No. 14, 67th Cong., 1st Sess., 7-8 (1921); H. R. Conf. Rep. No. 1044, 66th Cong., 2d Sess., 13 (1920); S. Rep. No. 524, 66th Cong., 2d Sess., 6-7 (1920); H. R. Rep. No. 362, 66th Cong., 1st Sess., 8-9 (1919). The debates over the Deficit Control Act contain no suggestion that the Comptroller General was chosen for the tasks outlined in §251 because Congress thought it could count on him to do its will; instead, the Comptroller General appears to have been selected precisely because of his independence from both the Legislature and the Executive. By assigning the reporting functions to the Comptroller General, rather than to the Congressional Budget Office or to the Office of Management and Budget, Congress sought to create “a wall . . . that takes these decisions out of the hands of the President and the Congress.” 131 Cong. Rec. 30865 (1985) (remarks of Rep. Gephardt) (emphasis added); see also, e. g., id., at 36089 (1985) (remarks of Rep. Weiss); id., at 36367 (1985) (remarks of Rep. Bedell).
Of course, the Deficit Control Act was hardly the first statute to assign new functions to the Comptroller General; a good number of other duties have been delegated to the Comptroller General over the years. But there is no reason to believe that, in effecting these earlier delegations, Congress relied any more heavily on the availability of the re*786moval provision than it did in passing the Deficit Control Act. In the past, as in 1985, it is far more likely that Congress was concerned mainly with the Comptroller General’s demonstrated political independence, and perhaps to a lesser extent with his long tradition of service to the Legislative Branch; neither of these characteristics depends to any significant extent on the ability of Congress to remove the Comptroller General without instituting impeachment proceedings. Striking down the congressional-removal provision might marginally frustrate the legislative expectations underlying some grants of authority to the Comptroller General, but surely to a lesser extent than would invalidation of § 251 of Gramm-Rudman-Hollings — along with all other “executive” powers delegated to the Comptroller General over the years.5
*787I do not claim that the 1921 removal provision is a piece of statutory deadwood utterly without contemporary significance. But it comes close. Rarely if ever invoked even for symbolic purposes, the removal provision certainly pales in importance beside the legislative scheme the Court strikes down today — an extraordinarily far-reaching response to a deficit problem of unprecedented proportions. Because I believe that the constitutional defect found by the Court cannot justify the remedy it has imposed, I respectfully dissent.
For the reasons identified by the District Court, I agree that the Deficit Control Act does not violate the nondelegation doctrine. See Synar v. United States, 626 F. Supp. 1374, 1382-1391 (DC 1986).
Justice Stevens concludes that the delegation effected under §251 contravenes the holding of INS v. Chadha, 462 U. S. 919 (1983), that Congress may make law only “in conformity with the express procedures of the Constitution’s prescription for legislative action: passage by a majority of both Houses and presentment to the President.” Id., at 958. I do not agree. We made clear in Chadha that the bicameralism and presentation requirements prevented Congress from itself exercising legislative power through some kind of procedural shortcut, such as the one-House veto challenged in that case. But we also made clear that our holding in no way questioned “Congress’ authority to delegate portions of its power to administrative agencies.” Id., at 953-954, n. 16. We explained: “Executive action under legislatively delegated authority that might resemble ‘legislative’ action in some respects is not subject to the approval of both Houses of Congress and the President for the reason that the Constitution does not so require. That kind of Executive action is always subject to check by the terms-of the legislation that authorized it; and if that authority is exceeded it is open to judicial review as well as the power of Congress to modify or revoke the authority entirely.” Ibid.
Although Justice Stevens seems to agree that the duties delegated to the Comptroller General under § 251 could be assigned constitutionally to an independent administrative agency, he argues that Congress may not give these duties “to one of its own agents.” Ante, at 752-753. He explains that the Comptroller General fits this description because “most” of his statutory responsibilities require him to provide services to Congress, and because Congress has repeatedly referred to the Comptroller General as part of the Legislative Branch. See ante, at 741-746. “If Congress were free to delegate its policymaking authority” to such an officer, Justice Stevens contends that “it would be able to evade ‘the carefully crafted restraints spelled out in the Constitution.’ ” Ante, at 755, quoting Chadha, 462 U. S., at 959. In his view, “[tjhat danger — congressional action that evades constitutional restraints — is not present when Congress delegates lawmaking power to the executive or to an independent agency.” Ante, at 755.
I do not think that danger is present here, either. The Comptroller General is not Congress, nor is he a part of Congress; “irrespective of Con*779gress’ designation,” he is an officer of the United States, appointed by the President. Buckley v. Valeo, 424 U. S. 1, 128, n. 165 (1976). In this respect the Comptroller General differs critically from, for example, the Director of the Congressional Budget Office, who is appointed by Congress, see 2 U. S. C. § 601(a)(2), and hence may not “exercis[e] significant authority pursuant to the laws of the United States,” Buckley v. Valeo, supra, at 126; see U. S. Const., Art. II, § 2, cl. 2. The exercise of rulemaking authority by an independent agency such as the Federal Trade Commission does not offend Chadha, even though the Commission could be described as an “agent” of Congress because it “carries] into effect legislative policies embodied in the statute in accordance with the legislative standard therein prescribed.” Humphrey’s Executor v. United States, 295 U. S. 602, 628 (1935). I do not see why the danger of “congressional action that evades constitutional restraints” becomes any more pronounced when a statute delegates power to a Presidentially appointed agent whose primary duties require him to provide services to Congress. The impermissibility of such a delegation surely is not rendered “obvious” by the fact that some officers who perform services for Congress have titles such as “librarian,” “architect,” or “printer.” See ante, at 758, n. 25 (Stevens, J., concurring in judgment). Furthermore, in sustaining the constitutionality of the Federal Trade Commission’s independent status, this Court noted specifically that the Commission “acts as a legislative agency” in “making investigations and reports thereon for the information of Congress ... in aid of the legislative power.” 295 U. S., at 628. Justice Stevens’ approach might make some sense if Congress had delegated legislative responsibility to an officer over whom Congress could hope to exercise tight control, but even Justice Stevens does not claim that the Comptroller General is such an officer.
In Myers v. United States, 272 U. S. 52 (1926), the Court refused to enforce a statute requiring congressional approval for removal of postmasters. The Court’s analysis suggested that there was no practical way the duties of the office could have been reformulated to render congressional participation in the removal process permissible. In Springer v. Philippine Islands, 277 U. S. 189 (1928), the Court removed from office several Philippine officials exercising executive powers but appointed by officers of the Philippine Legislature. As in Myers, the Court concluded that the offices by their very nature were executive, so the appointments *781could not have been rendered legal simply by trimming the delegated duties. In Buckley v. Valeo, 424 U. S. 1 (1976), the Court set aside Federal Election Campaign Act provisions granting certain powers to officials appointed by Congress, but it structured its remedy so as to interfere as little as possible with the orderly conduct of business by the Federal Election Commission. Past acts of the improperly constituted Commission were deemed valid, and the Court’s mandate was stayed for 30 days to allow time for the Commission to be reconstituted through Presidential appointment. See id,., at 142-143. Finally, in Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U. S. 50 (1982), the Court set aside an exercise of judicial power by a bankruptcy judge, because his tenure was not protected in the manner required by Article III of the Constitution. To give Article III protections to bankruptcy judges, the federal bankruptcy statute would have had to be rewritten completely.
Although the legislative history on this point is sparse, it seems reasonably clear that Congress intended the removal provision to be sever-able from the remainder of the 1921 statute. An earlier bill, providing for removal of the Comptroller General only by impeachment or concurrent resolution of Congress, was vetoed by President Wilson on the grounds that Congress could not constitutionally limit the President’s removal power or exercise such power on its own. See 59 Cong. Rec. 8609-8610 *784(1920). In the course of an unsuccessful attempt to override the veto, Representative Pell inquired: “If we pass this over the President’s veto and then the Supreme Court should uphold the contention of the President, this bill would not fail, would it? The bill would continue.” Representative Blanton answered, “Certainly.” Id., at 8611.
“All of the comptrollers general have treasured and defended the independence of their office, not alone from the president but also from the Congress itself. . . . Like the other institutions in the government, GAO depends upon Congress for its powers, its resources, and its general oversight. But it also possesses continuing legal powers, of both long and recent standing, that Congress has granted it and that it can exercise in a quite independent fashion. And the comptroller general, realistically speaking, is immune from removal during his fifteen-year term for anything short of a capital crime, a crippling illness, or insanity.” F. Mosher, A Tale of Two Agencies 158 (1984). See also, e. g., Ameron, Inc. v. United States Army Corps of Engineers, 787 F. 2d 875, 885-887 (CA3 1986); F. Mosher, The GAO 2, 240-244 (1979); H. Mansfield, The Comptroller General 75-76 (1939).
Many of the Comptroller General’s other duties, including those listed by the majority, see ante, at 734, n. 9, appear to meet the majority’s test for plainly “executive” functions— i. e., they require the Comptroller General to “[i]nterpre[t] a law enacted by Congress to implement the legislative mandate,” and to “exercise judgment concerning facts that affect the application of the [law].” Ante, at 733. Indeed, the majority’s approach would appear to classify as “executive” some of the most traditional duties of the Comptroller General, such as approving expenditure warrants, rendering conclusive decisions on the legality of proposed agency disbursements, and settling financial claims by and against the Government. See 31 U. S. C. §§ 3323, 3526-3529, 3702; F. Mosher, A Tale of Two Agencies 159-160 (1984). All three of these functions were given to the Comptroller General when the position was created in 1921. See 42 Stat. 20, 24-25.
I do not understand the majority’s assertion that invalidating the 1921 removal provision might make the Comptroller General “subservient to the Executive Branch.” Ante, at 734. The majority does not suggest that an official who exercises the functions that the Deficit Control Act vests in the Comptroller General must be removable by the President at will. Perhaps the President possesses inherent constitutional authority to remove “executive” officials for such politically neutral grounds as inefficiency or neglect of duty, but if so — and I am not convinced of it — I do not see how that power would be enhanced by nullification of a statutory provision giving similar authority to Congress. In any event, I agree with Justice White and Justice Stevens that the power to remove an officer for rea*787sons of this kind cannot realistically be expected to make an officer “subservient” in any meaningful sense to the removing authority. Cf. Humphrey’s Executor v. United States, 295 U. S., at 629.