These appeals present constitutional challenges to section 3 of the Defense of Marriage Act (“DOMA”), 1 U.S.C. § 7, which denies federal economic and other benefits to same-sex couples lawfully married in Massachusetts and to surviving spouses from couples thus married. Rather than challenging the right of states to define marriage as they see fit, the appeals contest the right of Congress to undercut the choices made by same-sex couples and by individual states in deciding who can be married to whom.
In 1993, the Hawaii Supreme Court held that it might violate the Hawaii constitution to deny marriage licenses to same-sex couples. Baehr v. Lewin, 74 Haw. 530, 852 P.2d 44, 48, 68 (1993). Although Ha*6waii then empowered its legislature to block such a ruling, Haw. Const, art. I, § 23 — which it did, Act of June 22, 1994, 1994 Haw. Sess. Laws 526 (H.B. 2312) (codified at Haw.Rev.Stat. § 572-1) — the Hawaii decision was followed by legalization of same-sex marriage in a small minority of states, some by statute and a few by judicial decision;1 many more states responded by banning same-sex marriage by statute or constitutional amendment.2
Congress reacted with the same alarm as many state legislatures. Within three years after the Hawaii decision, DOMA was enacted with strong majorities in both Houses and signed into law by President Clinton. The entire statute, reprinted in an addendum to this decision, must — having only two operative paragraphs — be one of the shortest major enactments in recent history. Section 3 of DOMA, 1 U.S.C. § 7, defines “marriage” for purposes of federal law:
In determining the meaning of any Act of Congress, or of any ruling, regulation, or interpretation of the various administrative bureaus and agencies of the United States, the word “marriage” means only a legal union between one man and one woman as husband and wife, and the word “spouse” refers only to a person of the opposite sex who is a husband or a wife.
Section 2, which is not at issue here, absolves states from recognizing same-sex marriages solemnized in other states.
DOMA does not formally invalidate same-sex marriages in states that permit them, but its adverse consequences for such a choice are considerable. Notably, it prevents same-sex married couples from filing joint federal tax returns, which can lessen tax burdens, see 26 U.S.C. § 1(a)-(c), and prevents the surviving spouse of a same-sex marriage from collecting Social Security survivor benefits, e.g., 42 U.S.C. § 402(f), (i). DOMA also leaves federal employees unable to share their health insurance and certain other medical benefits with same-sex spouses.
DOMA affects a thousand or more generic cross-references to marriage in myriad federal laws. In most cases, the changes operate to the disadvantage of same-sex married couples in the half dozen or so states that permit same-sex marriage. The number of couples thus affected is estimated at more than 100,000.3 Further, DOMA has potentially serious adverse consequences, hereafter described, for states that choose to legalize same-sex marriage.
In Gill v. OPM, No. 10-2207, seven same-sex couples married in Massachusetts and three surviving spouses of such marriages brought suit in federal district court to enjoin pertinent federal agencies *7and officials from enforcing DOMA to deprive the couples of federal benefits available to opposite-sex married couples in Massachusetts. The Commonwealth brought a companion case, Massachusetts v. DHHS, No. 10-2204, concerned that DOMA will revoke federal funding for programs tied to DOMA’s opposite-sex marriage definition — such as Massachusetts’ state Medicaid program and veterans’ cemeteries.
By combining the income of individuals in same-sex marriages, Massachusetts’ Medicaid program is noncompliant with DOMA, and the Department of Health and Human Services, through its Centers for Medicare and Medicaid Services, has discretion to rescind Medicaid funding to noncomplying states. Burying a veteran with his or her same-sex spouse removes federal “veterans’ cemetery” status and gives the Department of Veterans’ Affairs discretion to recapture all federal funding for the cemetery.
The Department of Justice defended DOMA in the district court but, on July 8, 2010, that court found section 3 unconstitutional under the Equal Protection Clause. Gill v. Office of Pers. Mgmt., 699 F.Supp.2d 374, 397 (D.Mass.2010). In the companion case, the district court accepted the Commonwealth’s argument that section 3 violated the Spending Clause and the Tenth Amendment. Massachusetts v. U.S. Dep’t of Health & Human Servs., 698 F.Supp.2d 234, 249, 253 (D.Mass.2010).
The district court’s judgment declared section 3 unconstitutional and enjoined the federal officials and agencies from enforcing section 3, but the court stayed injunctive relief pending appeals. The judgment included specific remedies ordered for the named plaintiffs in relation to tax, social security and like claims. With one qualification — discussed separately below — the federal defendants have throughout focused solely upon the district court’s premise that DOMA is unconstitutional.
The Justice Department filed a brief in this court defending DOMA against all constitutional claims. Thereafter, altering its position, the Justice Department filed a revised brief arguing that the equal protection claim should be assessed under a “heightened scrutiny” standard and that DOMA failed under that standard. It opposed the separate Spending Clause and Tenth Amendment claims pressed by the Commonwealth. The Gill plaintiffs defend the district court judgment on all three grounds.
A delay in proceedings followed the Justice Department’s about face while defense of the statute passed to a group of Republican leaders of the House of Representatives — the Bipartisan Legal Advisory Group (“the Legal Group”) — who retained counsel and intervened in the appeal to support section 3. A large number of amicus briefs have been filed on both sides of the dispute, some on both sides proving very helpful to the court.
On appeal from a grant of summary judgment, our review is de novo, Kuperman v. Wrenn, 645 F.3d 69, 73 (1st Cir. 2011), and the issues presented are themselves legal in character, even though informed by background information as to legislative purpose and “legislative facts” bearing upon the rationality or adequacy of distinctions drawn by statutes. E.g., FCC v. Beach Commc’ns, Inc., 508 U.S. 307, 314-20, 113 S.Ct. 2096, 124 L.Ed.2d 211 (1993). Such information is normally noticed by courts with the assistance of briefs, records and common knowledge. Daggett v. Comm’n on Governmental Ethics & Election Practices, 172 F.3d 104, 112 (1st Cir.1999).
This case is difficult because it couples issues of equal protection and federalism *8with the need to assess the rationale for a congressional statute passed with minimal hearings and lacking in formal findings. In addition, Supreme Court precedent offers some help to each side, but the rationale in several cases is open to interpretation. We have done our best to discern the direction of these precedents, but only the Supreme Court can finally decide this unique case.
Although our decision discusses equal protection and federalism concerns separately, it concludes that governing precedents under both heads combine — not to create some new category of “heightened scrutiny” for DOMA under a prescribed algorithm, but rather to require a closer than usual review based in part on discrepant impact among married couples and in part on the importance of state interests in regulating marriage. Our decision then tests the rationales offered for DOMA, taking account of Supreme Court precedent limiting which rationales can be counted and of the force of certain rationales.
Equal Protection. The Legal Group says that any equal protection challenge to DOMA is foreclosed at the outset by Baker v. Nelson, 409 U.S. 810, 93 S.Ct. 37, 34 L.Ed.2d 65 (1972). There, a central claim made was that a state’s refusal to recognize same-sex marriage violated federal equal protection principles. Minnesota had, like DOMA, defined marriage as a union of persons of the opposite sex, and the state supreme court had upheld the statute. On appeal, the Supreme Court dismissed summarily for want of a substantial federal question. Id.
Baker is precedent binding on us unless repudiated by subsequent Supreme Court precedent. Hicks v. Miranda, 422 U.S. 332, 344, 95 S.Ct. 2281, 45 L.Ed.2d 223 (1975). Following Baker, “gay rights” claims prevailed in several well known decisions, Lawrence v. Texas, 539 U.S. 558, 123 S.Ct. 2472, 156 L.Ed.2d 508 (2003), and Romer v. Evans, 517 U.S. 620, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996),4 but neither mandates that the Constitution requires states to permit same-sex marriages. A Supreme Court summary dismissal “prevents] lower courts from coming to opposite conclusions on the precise issues presented and necessarily decided by those actions.” Mandel v. Bradley, 432 U.S. 173, 176, 97 S.Ct. 2238, 53 L.Ed.2d 199 (1977) (per curiam). Baker does not resolve our own case but it does limit the arguments to ones that do not presume or rest on a constitutional right to same-sex marriage.
Central to this appeal is Supreme Court case law governing equal protection analysis. The Gill plaintiffs say that DOMA fails under the so-called rational basis test, traditionally used in cases not involving “suspect” classifications. The federal defendants said that DOMA would survive such rational basis scrutiny but now urge, instead, that DOMA fails under so-called intermediate scrutiny. In our view, these competing formulas are inadequate fully to describe governing precedent.
Certain suspect classifications— race, alienage and national origin — require what the Court calls strict scrutiny, which entails both a compelling governmental in-*9terest and narrow tailoring. Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 227, 115 S.Ct. 2097, 132 L.Ed.2d 158 (1995). Gender-based classifications invoke intermediate scrutiny and must be substantially related to achieving an important governmental objective.5 Both are far more demanding than rational basis review as conventionally applied in routine matters of commercial, tax and like regulation.
Equal protection claims tested by this rational basis standard, famously called by Justice Holmes the “last resort of constitutional argument,” Buck v. Bell, 274 U.S. 200, 208, 47 S.Ct. 584, 71 L.Ed. 1000 (1927), rarely succeed. Courts accept as adequate any plausible factual basis, Williamson v. Lee Optical of Oklahoma, Inc., 348 U.S. 483, 487-88, 75 S.Ct. 461, 99 L.Ed. 563 (1955), without regard to Congress’ actual motives. Beach Commc’ns, 508 U.S. at 314, 113 S.Ct. 2096. Means need not be narrowly drawn to meet — or even be entirely consistent with — the stated legislative ends. Lee Optical, 348 U.S. at 487-88, 75 S.Ct. 461.
Under such a rational basis standard, the Gill plaintiffs cannot prevail. Consider only one of the several justifications for DOMA offered by Congress itself, namely, that broadening the definition of marriage will reduce tax revenues and increase social security payments. This is the converse of the very advantages that the Gill plaintiffs are seeking, and Congress could rationally have believed that DOMA would reduce costs, even if newer studies of the actual economic effects of DOMA suggest that it may in fact raise costs for the federal government.
The federal defendants conceded that rational basis review leaves DOMA intact but now urge this court to employ the so-called intermediate scrutiny test used by Supreme Court for gender discrimination. Some similarity exists between the two situations along with some differences, compare Frontiero v. Richardson, 411 U.S. 677, 682-88, 93 S.Ct. 1764, 36 L.Ed.2d 583 (1973) (plurality opinion) (describing criteria for categorization). But extending intermediate scrutiny to sexual preference classifications is not a step open to us.
First, this court in Cook v. Gates, 528 F.3d 42 (1st Cir.2008), cert. denied, — U.S.-, 129 S.Ct. 2763, 174 L.Ed.2d 284 (2009), has already declined to create a major new category of “suspect classification” for statutes distinguishing based on sexual preference. Cook rejected an equal protection challenge to the now-superceded “Don’t Ask, Don’t Tell” policy adopted by Congress for the military, pointing out that Romer itself avoided the suspect classification label. Cook, 528 F.3d at 61-62. This binds the panel. San Juan Cable LLC v. P.R. Tel. Co., 612 F.3d 25, 33 (1st Cir.2010).
Second, to create such a new suspect classification for same-sex relationships would have far-reaching implications — in particular, by implying an overruling of Baker, which we are neither empowered to do nor willing to predict. Nothing indicates that the Supreme Court is about to adopt this new suspect classification when it conspicuously failed to do so in Romer— a case that could readily have been disposed by such a demarche. That such a *10classification could overturn marriage laws in a huge majority of individual states underscores the implications.
However, that is not the end of the matter. Without relying on suspect classifications, Supreme Court equal protection decisions have both intensified scrutiny of purported justifications where minorities are subject to discrepant treatment and have limited the permissible justifications. And (as we later explain), in areas where state regulation has traditionally governed, the Court may require that the federal government interest in intervention be shown with special clarity.
In a set of equal protection decisions, the Supreme Court has now several times struck down state or local enactments without invoking any suspect classification. In each, the protesting group was historically disadvantaged or unpopular, and the statutory justification seemed thin, unsupported or impermissible. It is these decisions — not classic rational basis review— that the Gill plaintiffs and the Justice Department most usefully invoke in their briefs (while seeking to absorb them into different and more rigid categorical rubrics).
The oldest of the decisions, U.S. Dept. of Agric. v. Moreno, 413 U.S. 528, 93 S.Ct. 2821, 37 L.Ed.2d 782 (1973), invalidated Congress’ decision to exclude from the food stamp program households containing unrelated individuals. Disregarding purported justifications that such households were more likely to under-report income and to evade detection, the Court closely scrutinized the legislation’s fit — finding both that the rule disqualified many otherwise-eligible and particularly needy households, and a “bare congressional desire to harm a politically unpopular group.” Id. at 534, 537-38, 93 S.Ct. 2821.
The second, City of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 105 S.Ct. 3249, 87 L.Ed.2d 313 (1985), overturned a local ordinance as applied to the denial of a special permit for operating a group home for the mentally disabled. The Court found unconvincing interests like protecting the inhabitants against the risk of flooding, given that nursing or convalescent homes were allowed without a permit; mental disability too had no connection to alleged concerns about population density. All that remained were “mere negative attitudes, or fear, unsubstantiated by factors which are properly cognizable in a zoning proceeding.” Id. at 448, 105 S.Ct. 3249.
Finally, in Romer v. Evans, 517 U.S. 620, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996), the Court struck down a provision in Colorado’s constitution prohibiting regulation to protect homosexuals from discrimination. The Court, calling “unprecedented” the “disqualification of a class of persons from the right to seek specific protection from the law,” deemed the provision a “status-based enactment divorced from any factual context from which we could discern a relationship to legitimate state interests.” Id. at 632-33, 635, 116 S.Ct. 1620.
These three decisions did not adopt some new category of suspect classification or employ rational basis review in its minimalist form; instead, the Court rested on the case-specific nature of the discrepant treatment, the burden imposed, and the infirmities of the justifications offered. Several Justices have remarked on this— both favorably, City of Cleburne, 473 U.S. at 451-55, 105 S.Ct. 3249 (1985) (Stevens, J., concurring), and unfavorably, United States v. Virginia (VMI), 518 U.S. 515, 567, 116 S.Ct. 2264, 135 L.Ed.2d 735 (1996) (Scalia, J., dissenting).
Circuit courts, citing these same cases, have similarly concluded that equal protec*11tion assessments are sensitive to the circumstances of the case and not dependent entirely on abstract categorizations.6 As one distinguished judge observed:
Judges and commentators have noted that the usually deferential “rational basis” test has been applied with greater rigor in some contexts, particularly those in which courts have had reason to be concerned about possible discrimination.
United States v. Then, 56 F.3d 464, 468 (2d Cir.1995) (Calabresi, J., concurring) (citing City of Cleburne as an example). There is nothing remarkable about this: categories are often approximations and are themselves constructed by weighing of underlying elements.
All three of the cited cases — Moreno, City of Cleburne and Romer — stressed the historic patterns of disadvantage suffered by the group adversely affected by the statute. As with the women, the poor and the mentally impaired, gays and lesbians have long been the subject of discrimination. Lawrence, 539 U.S. at 571, 123 S.Ct. 2472. The Court has in these cases undertaken a more careful assessment of the justifications than the light scrutiny offered by conventional rational basis review.
As for burden, the combined effect of DOMA’s restrictions on federal benefits will not prevent same-sex marriage where permitted under state law; but it will penalize those couples by limiting tax and social security benefits to opposite-sex couples in their own and all other states. For those married same-sex couples of which one partner is in federal service, the other cannot take advantage of medical care and other benefits available to opposite-sex partners in Massachusetts and everywhere else in the country.
These burdens are comparable to those the Court found substantial in Moreno, City of Cleburne, and Romer. Moreno, like this case, involved meaningful economic benefits; City of Cleburne involved the opportunity to secure housing; Romer, the chance to secure equal protection of the laws on the same terms as other groups. Loss of survivor’s social security, spouse-based medical care and tax benefits are major detriments on any reckoning; provision for retirement and medical care are, in practice, the main components of the social safety net for vast numbers of Americans.
Accordingly, we conclude that the extreme deference accorded to ordinary economic legislation in cases like Lee Optical would not be extended to DOMA by the Supreme Court; and without insisting on “compelling” or “important” justifications or “narrow tailoring,” the Court would scrutinize with care the purported bases for the legislation. Before providing such scrutiny, a separate element absent in Moreno, City of Cleburne, and Romer— federalism — must be considered.
Federalism. In assailing DOMA, the plaintiffs and especially the Commonwealth rely directly on limitations attributed to the Spending Clause of the Constitution and the Tenth Amendment; the Justice Department, along with the Legal Group, rejects those claims. In our view, neither the Tenth Amendment nor the Spending Clause invalidates DOMA; but Supreme Court precedent relating to federalism-based challenges to federal laws reinforce the need for closer than usual *12scrutiny of DOMA’s justifications and diminish somewhat the deference ordinarily accorded.
It is true that DOMA intrudes extensively into a realm that has from the start of the nation been primarily confided to state regulation — domestic relations and the definition and incidents of lawful marriage — which is a leading instance of the states’ exercise of their broad police-power authority over morality and culture. As the Supreme Court observed long ago,
[t]he whole subject of the domestic relations of husband and wife, parent and child, belongs to the laws of the States and not to the laws of the United States.
Hisquierdo v. Hisquierdo, 439 U.S. 572, 581, 99 S.Ct. 802, 59 L.Ed.2d 1 (1979) (quoting In re Burrus, 136 U.S. 586, 593-94, 10 S.Ct. 850, 34 L.Ed. 500 (1890)); see also Loving v. Virginia, 388 U.S. 1, 7, 87 S.Ct. 1817, 18 L.Ed.2d 1010 (1967) (marriage).
Consonantly, Congress has never purported to lay down a general code defining marriage or purporting to bind the states to such a regime. Rather, in individual situations — such as the anti-fraud criteria in immigration law, 8 U.S.C. § 1186a(b)(l)(A)(i) — Congress has provided its own definitions limited to the particular program or personnel involved. But no precedent exists for DOMA’s sweeping general “federal” definition of marriage for all federal statutes and programs.
Nevertheless, Congress surely has an interest in who counts as married. The statutes and programs that section 3 governs are federal regimes such as social security, the Internal Revenue Code and medical insurance for federal workers; and their benefit structure requires deciding who is married to whom. That Congress has traditionally looked to state law to determine the answer does not mean that the Tenth Amendment or Spending Clause require it to do so.
Supreme Court interpretations of the Tenth Amendment have varied over the years but those in force today have struck down statutes only where Congress sought to commandeer state governments or otherwise directly dictate the internal operations of state government. Printz v. United States, 521 U.S. 898, 935, 117 S.Ct. 2365, 138 L.Ed.2d 914 (1997); New York v. United States, 505 U.S. 144, 188, 112 S.Ct. 2408, 120 L.Ed.2d 120 (1992). Whatever its spin-off effects, section 3 governs only federal programs and funding, and does not share these two vices of commandeering or direct command.
Neither does DOMA run afoul of the “germaneness” requirement that conditions on federal funds must be related to federal purposes. South Dakota v. Dole, 483 U.S. 203, 207-08, 107 S.Ct. 2793, 97 L.Ed.2d 171 (1987). The requirement is not implicated where, as here, Congress merely defines the terms of the federal benefit. In Dole, the Supreme Court upheld a condition by which federal funds for highway construction depended on a state’s adoption of a minimum drinking age for all driving on state roadways. 483 U.S. at 205, 107 S.Ct. 2793. DOMA merely limits the use of federal funds to prescribed purposes.
However, the denial of federal benefits to same-sex couples lawfully married does burden the choice of states like Massachusetts to regulate the rules and incidents of marriage; notably, the Commonwealth stands both to assume new administrative burdens and to lose funding for Medicaid or veterans’ cemeteries solely on account of its same-sex marriage laws. These consequences do not violate the Tenth Amendment or Spending Clause, but Congress’ effort to put a thumb on the scales and influence a state’s decision as to how to *13shape its own marriage laws does bear on how the justifications are assessed.
In United States v. Morrison, 529 U.S. 598, 120 S.Ct. 1740, 146 L.Ed.2d 658 (2000), and United States v. Lopez, 514 U.S. 549, 115 S.Ct. 1624, 131 L.Ed.2d 626 (1995), the Supreme Court scrutinized with special care federal statutes intruding on matters customarily within state control. The lack of adequate and persuasive findings led the Court in both cases to invalidate the statutes under the Commerce Clause even though nothing more than rational basis review is normally afforded in such cases.
The Supreme Court has made somewhat similar statements about the need for scrutiny when examining federal statutes intruding on regulation of state election processes. Nw. Austin Mun. Util. Dist. No. One v. Holder, 557 U.S. 193, 129 S.Ct. 2504, 2511, 174 L.Ed.2d 140 (2009);7 cf. City of Boerne v. Flores, 521 U.S. 507, 534, 117 S.Ct. 2157, 138 L.Ed.2d 624 (1997) (calling RFRA a “considerable congressional intrusion into the States’ traditional prerogatives and general authority to regulate for the health and welfare of their citizens”).
True, these federalism cases examined the reach of federal power under the Commerce Clause and other sources of constitutional authority not invoked here; but a statute that violates equal protection is likewise beyond the power of Congress. See Moreno, 413 U.S. at 541, 93 S.Ct. 2821 (Douglas, J., concurring). Given that DOMA intrudes broadly into an area of traditional state regulation, a closer examination of the justifications that would prevent DOMA from violating equal protection (and thus from exceeding federal authority) is uniquely reinforced by federalism concerns.
DOMA’s Rationales. Despite its ramifying application throughout the U.S. Code, only one day of hearings was held on DOMA, Defense of Marriage Act: Hearing on H.R. 3396 Before the Subcomm. on the Constitution of the H. Comm, on the Judiciary, 104th Cong. (1996) (“Hearing”), and none of the testimony concerned DOMA’s effects on the numerous federal programs at issue. Some of the odder consequences of DOMA testify to the speed with which it was adopted.8
The statute, only a few paragraphs in length, is devoid of the express prefatory findings commonly made in major federal laws. E.g., 15 U.S.C. § 80a-l; 16 U.S.C. § 1531; 20 U.S.C. § 1400; 21 U.S.C. § 801; 29 U.S.C. § 151; id. § 1001; 42 *14U.S.C. § 7401. Accordingly, in discerning and assessing Congress’ basis for DOMA our main resort is the House Committee report and, in lesser measure, to variations of its themes advanced in the briefs before us. The committee report stated:
[T]he Committee briefly discusses four of the governmental interests advanced by this legislation: (1) defending and nurturing the institution of traditional, heterosexual marriage; (2) defending traditional notions of morality; (3) protecting state sovereignty and democratic self-governance; and (4) preserving scarce government resources.
H.R.Rep. No. 104-664, at 12 (1996), 1996 U.S.C.C.A.N. 2905, 2916.
The penultimate reason listed above was not directed to section 3 — indeed, is antithetical to it — but was concerned solely with section 2, which reserved a state’s power not to recognize same-sex marriages performed in other states. Thus, we begin with the others, reserving for separate consideration the claim strongly pressed by the Gill plaintiffs that DOMA should be condemned because its unacknowledged but alleged central motive was hostility to homosexuality.
First, starting with the most concrete of the cited reasons — “preserving scarce government resources” — it is said that DOMA will save money for the federal government by limiting tax savings and avoiding social security and other payments to spouses. This may well be true, or at least might have been thought true; more detailed recent analysis indicates that DOMA is more likely on a net basis to cost the government money.9
But, where the distinction is drawn against a historically disadvantaged group and has no other basis, Supreme Court precedent marks this as a reason undermining rather than bolstering the distinction. Plyler v. Doe, 457 U.S. 202, 227, 102 S.Ct. 2382, 72 L.Ed.2d 786 (1982); Romer, 517 U.S. at 635, 116 S.Ct. 1620. The reason, derived from equal protection analysis, is that such a group has historically been less able to protect itself through the political process. Plyler, 457 U.S. at 218 n. 14, 102 S.Ct. 2382; United States v. Carolene Prods. Co., 304 U.S. 144, 152 n. 4, 58 S.Ct. 778, 82 L.Ed. 1234 (1938).
A second rationale of a pragmatic character, advanced by the Legal Group’s brief and several others, is to support child-rearing in the context of stable marriage.10 The evidence as to child rearing by same-sex couples is the subject of controversy, but we need not enter the debate. Whether or not children raised by opposite-sex marriages are on average better served, DOMA cannot preclude same-sex couples in Massachusetts from adopting children or prevent a woman partner from giving birth to a child to be raised by both partners.
Although the House Report is filled with encomia to heterosexual marriage, DOMA does not increase benefits to opposite-sex couples — whose marriages may in any event be childless, unstable or both — or explain how denying benefits to same-sex couples will reinforce heterosexual marriage. Certainly, the denial will not affect *15the gender choices of those seeking marriage. This is not merely a matter of poor fit of remedy to perceived problem, Lee Optical, 348 U.S. at 487-88, 75 S.Ct. 461; City of Cleburne, 473 U.S. at 446-50, 105 S.Ct. 3249, but a lack of any demonstrated connection between DOMA’s treatment of same-sex couples and its asserted goal of strengthening the bonds and benefits to society of heterosexual marriage.
A third reason, moral disapproval of homosexuality, is one of DOMA’s stated justifications:
Civil laws that permit only heterosexual marriage reflect and honor a collective moral judgment about human sexuality. This judgment entails both moral disapproval of homosexuality, and a moral conviction that heterosexuality better comports with traditional (especially Judeo-Christian) morality.
H.R.Rep. No. 104-664, at 15-16 (emphasis added); see also, e.g., 142 Cong. Rec. 16,-972 (1996) (statement of Rep. Coburn) (homosexuality “morally wrong”).
For generations, moral disapproval has been taken as an adequate basis for legislation, although usually in choices made by state legislators to whom general police power is entrusted. But, speaking directly of same-sex preferences, Lawrence ruled that moral disapproval alone cannot justify legislation discriminating on this basis. 539 U.S. at 577-78, 123 S.Ct. 2472. Moral judgments can hardly be avoided in legislation, but Lawrence and Romer have undercut this basis. Cf. Palmore v. Sidoti, 466 U.S. 429, 433, 104 S.Ct. 1879, 80 L.Ed.2d 421 (1984).
Finally, it has been suggested by the Legal Group’s brief that, faced with a prospective change in state marriage laws, Congress was entitled to “freeze” the situation and reflect. But the statute was not framed as a temporary time-out; and it has no expiration date, such as one that Congress included in the Voting Rights Act. See Nw. Austin, 129 S.Ct. at 2510 (describing original expiration date and later extensions); City of Boerne, 521 U.S. at 533, 117 S.Ct. 2157. The House Report’s own arguments — moral, prudential and fiscal — make clear that DOMA was not framed as a temporary measure.
Congress did emphasize a related concern, based on the Hawaii Supreme Court’s decision in Baehr, that state judges would impose same-sex marriage on unwilling states. H.R.Rep. No. 104-664, at 5-6, 12, 16-17. But almost all states have readily amended constitutions, as well as elected judges, and can protect themselves against what their citizens may regard as overreaching. The fear that Hawaii could impose same-sex marriage on sister states through the Full Faith and Credit Clause, id. at 7-9, relates solely to section 2 of DOMA, which is not before us.
We conclude, without resort to suspect classifications or any impairment of Baker, that the rationales offered do not provide adequate support for section 3 of DOMA. Several of the reasons given do not match the statute and several others are diminished by specific holdings in Supreme Court decisions more or less directly on point. If we are right in thinking that disparate impact on minority interests and federalism concerns both require somewhat more in this case than almost automatic deference to Congress’ will, this statute fails that test.
Invalidating a federal statute is an unwelcome responsibility for federal judges; the elected Congress speaks for the entire nation, its judgment and good faith being entitled to utmost respect. Gregg v. Georgia, 428 U.S. 153, 175, 96 S.Ct. 2909, 49 L.Ed.2d 859 (1976) (plurality opinion). But a lower federal court such as ours must follow its best understanding of gov*16erning precedent, knowing that in large matters the Supreme Court will correct mis-readings (and even if it approves the result will formulate its own explanation).
In reaching our judgment, we do not rely upon the charge that DOMA’s hidden but dominant purpose was hostility to homosexuality. The many legislators who supported DOMA acted from a variety of motives, one central and expressed aim being to preserve the heritage of marriage as traditionally defined over centuries of Western civilization. See H.R.Rep. No. 104-664, at 12,16. Preserving this institution is not the same as “mere moral disapproval of an excluded group,” Lawrence, 539 U.S. at 585, 123 S.Ct. 2472 (O’Connor, J., concurring), and that is singularly so in this case given the range of bipartisan support for the statute.
The opponents of section 3 point to selected comments from a few individual legislators; but the motives of a small group cannot taint a statute supported by large majorities in both Houses and signed by President Clinton. Traditions are the glue that holds society together, and many of our own traditions rest largely on belief and familiarity — not on benefits firmly provable in court. The desire to retain them is strong and can be honestly held.
For 150 years, this desire to maintain tradition would alone have been justification enough for almost any statute. This judicial deference has a distinguished lineage, including such figures as Justice Holmes, the second Justice Harlan, and Judges Learned Hand and Henry Friendly. But Supreme Court decisions in the last fifty years call for closer scrutiny of government action touching upon minority group interests and of federal action in areas of traditional state concern.
To conclude, many Americans believe that marriage is the union of a man and a woman, and most Americans live in states where that is the law today. One virtue of federalism is that it permits this diversity of governance based on local choice, but this applies as well to the states that have chosen to legalize same-sex marriage. Under current Supreme Court authority, Congress’ denial of federal benefits to same-sex couples lawfully married in Massachusetts has not been adequately supported by any permissible federal interest.
Hara’s Health Benefits Claim. A distinct, if much narrower, issue is raised by Dean Hara, one of the Gill plaintiffs. Although the district court ordered the relief Hara sought for Social Security lump-sum death benefits, the district court found that relief on his second claim for health coverage required a further determination on a precondition that is the subject of a proceeding earlier brought by Hara and now pending in the Federal Circuit. Hara v. Office of Pers. Mgmt., No. 2009-3134 (Fed. Cir. docketed Mar. 17, 2009).
Hara was married under Massachusetts law to a now-deceased Congressman, and Hara has sought to be enrolled as a surviving spouse for health benefits under the Congressman’s Federal Employees’ Health Benefit Plan (“FEHBP”). For this, (1) Hara would have to be an eligible “annuitant” under the annuity statute, and (2) the Congressman had to have enrolled in the health benefit plan for “self and family,” which he had not done. 5 U.S.C. § 8341; 5 C.F.R. §§ 890.303(c), 890.302(a)(1).
Acting on an application by Hara for a survivor annuity benefit, the Office of Personnel Management (“OPM”) had previously ruled that Hara was ineligible to receive an annuity both because he was not a spouse under DOMA and because the Congressman had not elected such coverage. Such determinations as to annuities are reviewed exclusively by the Merit *17Systems Protection Board (“MSPB” or “Board”) and then exclusively by the Federal Circuit. 5 U.S.C. §§ 8347, 8341, 7703(b)(1); 28 U.S.C. 1295(a)(9).
On review, the Board upheld the denial of coverage solely because of DOMA, finding the failure to elect coverage not to bar annuitant status. Hara sought further review in the Federal Circuit, and that case has been stayed pending resolution of the DOMA issue in this circuit. Hara, No. 2009-3134 (Oct. 15, 2010 order staying proceedings). Thus, now — as at the time the distinct court issued its judgment — a Board determination is in force that Hara lacks annuitant status.
OPM has separately denied Hara’s claim for FEHBP health enrollment because of the Congressman’s failure to elect “self and family” coverage. Although the district court found DOMA unconstitutional, it refused to resolve Hara’s health coverage claim now because it still depends on Hara establishing eligibility for annuitant status, which is at issue in his pending Federal Circuit appeal. Whether or not Hara lacked standing, the district court showed prudence in deferring on this issue to the Federal Circuit.
Hara says in substance that the Federal Circuit has to recognize his annuitant status because the Board has waived or forfeited any objection based on the failure to elect spousal survivor coverage; but the Department of Justice does not concede the point, which the Federal Circuit presumably will resolve. If Hara prevails there, district court injunctive relief to secure his health coverage is likely to be unnecessary, but our affirmance is without prejudice to such a future request by Hara.
The judgment of the district court is affirmed for the reasons and to the extent stated above. Anticipating that certiorari will be sought and that Supreme Court review of DOMA is highly likely, the mandate is stayed, maintaining the district court’s stay of its injunctive judgment, pending further order of this court. The parties will bear their own costs on these appeals.
It is so ordered.
ADDENDUM
SECTION. 1. SHORT TITLE.
This Act may be cited as the “Defense of Marriage Act”.
SEC. 2. POWERS RESERVED TO THE STATES.
(a) IN GENERAL. — Chapter 115 of title 28, United States Code, is amended by adding after section 1738B the following:
“§ 1738C. Certain acts, records, and proceedings and the effect thereof.”.
“No State, territory, or possession of the United States, or Indian tribe, shall be required to give effect to any public act, record, or judicial proceeding of any other State, territory, possession, or tribe respecting a relationship between persons of the same sex that is treated as a marriage under the laws of such other State, territory, possession, or tribe, or a right or claim arising from such relationship.”.
(b) CLERICAL AMENDMENT. — The table of sections at the beginning of chapter 115 of title 28, United States Code, is amended by inserting after the item relating to section 1738B the following new item:
“1738C. Certain acts, records, and proceedings and the effect thereof.”.
SEC. 3. DEFINITION OF MARRIAGE.
(a) IN GENERAL. — Chapter 1 of title 1, United States Code, is amended by adding at the end the following:
*18“§ 7. Definition of ‘marriage’ and ‘spouse’.”
“In determining the meaning of any Act of Congress, or of any ruling, regulation, or interpretation of the various administrative bureaus and agencies of the United States, the word ‘marriage’ means only a legal union between one man and one woman as husband and wife, and the word ‘spouse’ refers only to a person of the opposite sex who is a husband or a wife.”.
(b) CLERICAL AMENDMENT. — The table of sections at the beginning of chapter 1 of title 1, United States Code, is amended by inserting after the item relating to section 6 the following new item:
“7. Definition of ‘marriage’ and ‘spouse’.”
. E.g., Marriage Equality Act, 2011 N.Y. Sess. Laws. ch. 95 (A. 8354) (McKinney) (codified at N.Y. Dom. Rel. Law § 10 — a); Act of Feb. 13, 2012, 2012 Wash. Legis. Serv. ch. 3 (S.S.B. 6239) (West); Varnum v. Brien, 763 N.W.2d 862 (Iowa 2009); Goodridge v. Dep’t of Pub. Health, 440 Mass. 309, 798 N.E.2d 941, 969 (2003).
. E.g., Tex. Const, art. 1, § 32; Va Const, art. I, § 15-A; Act of May 24, 1996, 1996 Ill. Legis. Serv. P.A. 89-459 (S.B.1140) (West) (codified at 750 Ill. Comp. Stat. 5/212(a)(5)); Act of May 13, 1997, 1997 Ind. Legis. Serv. P.L. 198-1997 (H.E.A. 1265) (West) (codified at Ind.Code § 31-11-1-1).
.U.S. Census Bureau, Census Bureau Releases Estimates of Same-Sex Married Couples (Sept. 27, 2011), http://www.census.gov/ newsroom/releases/archives/2010_census/cb ll-cnl81.html; U.S. Census Bureau, Same-Sex Unmairied Partner or Spouse Households by Sex of Householder by Presence of Own Children: 2010 Census and 2010 American Community Smvey, http://www.census.gov/ hhes/samesex/files/supp-table-AFF.xls (last visited May 22, 2012).
. Lawrence struck down. Texas' statute forbidding homosexual sodomy and Romer overturned a Colorado constitutional amendment that curtailed the right of communities to enact laws to prevent discrimination against gays and lesbians. Although Lawrence rested on substantive due process precedent and not equal protection, precedents under the two rubrics use somewhat related tests as to levels of scrutiny — applied to liberty interests under the former and discrimination claims under the latter. Lawrence, 539 U.S. at 575-76, 578, 123 S.Ct. 2472; Romer, 517 U.S. at 632, 635, 116 S.Ct. 1620.
. United States v. Virginia (VMI), 518 U.S. 515, 532-33, 116 S.Ct. 2264, 135 L.Ed.2d 735 (1996); Mississippi Univ. for Women v. Hogan, 458 U.S. 718, 724, 102 S.Ct. 3331, 73 L.Ed.2d 1090 (1982); Craig v. Boren, 429 U.S. 190, 197, 97 S.Ct. 451, 50 L.Ed.2d 397 (1976); Frontiero v. Richardson, 411 U.S. 677, 682, 93 S.Ct. 1764, 36 L.Ed.2d 583 (1973) (plurality opinion).
. E.g., Ramos v. Town of Vernon, 353 F.3d 171, 175 (2d Cir.2003); Milner v. Apfel, 148 F.3d 812, 816 (7th Cir.), cert. denied, 525 U.S. 1024, 119 S.Ct. 556, 142 L.Ed.2d 463 (1998); Price v. Tanner, 855 F.2d 820, 829 n. 4 (11th Cir. 1988), cert. denied, 489 U.S. 1081, 109 S.Ct. 1534, 103 L.Ed.2d 839 (1989).
. The majority, focusing on the related issue of fit, said that "a departure from the fundamental principle of equal sovereignty [between states] requires a showing that a statute's disparate geographic coverage is sufficiently related to the problem that it targets.” Nw. Austin, 129 S.Ct. at 2512. Justice Thomas went a step further, stating “because States still retain sovereign authority over their election systems, any measure enacted in furtherance of the Fifteenth Amendment must be closely examined to ensure that its encroachment on state authority in this area is limited to the appropriate enforcement of this ban on discrimination.” Id. at 2520 (Thomas, J., concurring in part and dissenting in part).
. For example, DOMA's definition of marriage arguably undermines both federal ethics laws, 5 U.S.C. app. §§ 102(e)(l)(A)-(D), 501(c), and abuse reporting requirements in the military, 10 U.S.C. § 1787(a), insofar as it facially excludes same-sex married couples from their strictures. Other curiosities likely unintended are possible impacts on anti-nepotism provisions, 5 U.S.C. §§ 3110(a)(3), (b), 2302(b)(7); judicial recusals, 28 U.S.C. § 455(b)(4), restrictions on receipt of gifts, 2 U.S.C. § 31-2(a), and on travel reimbursement, 31 U.S.C. § 1353(a); and the crimes of bribery of federal officials, 18 U.S.C. § 208(a), and threats to family members of federal officials, id. § 115.
. Cong. Budget Office, The Potential Budgetary Impact of Recognizing Same-Sex Marriages (2004), available at http://www.cbo.gov/ sites/default/files/cbofiles/ftpdocs/55xx/doc 555 9/06-21 -samesexmarriage.pdf.
. The House Report refers obliquely to the importance of heterosexual marriage in "encouraging responsible procreation and child-rearing,” H.R.Rep. No. 104-664, at 13, but the subcommittee chair at the House hearing began by saying that "heterosexual marriage provides the ideal structure within which to beget and raise children.” Hearing, supra, at 1 (opening statement of Rep. Canady).