dissenting.
The Court today upholds a reapportionment scheme for a state legislature featuring an 89% maximum deviation and a 16% average deviation from population equality. I cannot agree.
I
Although I disagree with today’s holding, it is worth stressing how extraordinarily narrow it is, and how empty of likely precedential value. The Court goes out of its way to make clear that because appellants have chosen to attack only one small feature of Wyoming’s reapportionment scheme, the Court weighs only the marginal unequalizing effect of that one feature, and not the overall constitutionality of the entire scheme. Ante, at 846, and nn. 8, 9; see ante, *851at 849 (O’Connor, J., concurring). Hence, although in my view the Court reaches the wrong result in the case at hand, it is unlikely that any future plaintiffs challenging a state reapportionment scheme as unconstitutional will be so unwise as to limit their challenge to the scheme’s single most objectionable feature. Whether this will be a good thing for the speed and cost of constitutional litigation remains to be seen. But at least plaintiffs henceforth will know better than to exercise moderation or restraint in mounting constitutional attacks on state apportionment statutes, lest they forfeit their small claim by omitting to assert a big one.
B
t>
The Equal Protection Clause of the Fourteenth Amendment requires that a State, in apportioning its legislature, “make an honest and good faith effort to construct districts . . . as nearly of equal population as is practicable.” Reynolds v. Sims, 377 U. S. 533, 577 (1964). Under certain conditions the Constitution permits small deviations from absolute equality in state legislative districts,1 but we have carefully circumscribed the range of permissible deviations as to both degree and kind. What is required is “a faithful adherence to a plan of population-based representation, with such minor deviations only as may occur in recognizing certain factors that are free from any taint of arbitrariness or discrimination.” Roman v. Sincock, 377 U. S. 695, 710 (1964). “[T]he overriding objective must be substantial equality of population among the various districts, so that the vote of any citizen is approximately equal in weight to that of any other citizen in the State.” Reynolds, supra, at 579.
*852Our cases since Reynolds have clarified the structure of constitutional inquiry into state legislative apportionments, setting up what amounts to a four-step test. First, a plaintiff must show that the deviations at issue are sufficiently large to make out a prima facie case of discrimination. We have come to establish a rough threshold of 10% maximum deviation from equality (adding together the deviations from average district size of the most underrepresented and most overrepresented districts); below that level, deviations will ordinarily be considered de minimis. Ante, at 842-843; Connor v. Finch, 431 U. S. 407, 418 (1977); White v. Regester, 412 U. S. 755, 763-764 (1973). Second, a court must consider the quality of the reasons advanced by the State to explain the deviations. Acceptable reasons must be “legitimate considerations incident to the effectuation of a rational state policy,” Reynolds, supra, at 579, and must be “free from any taint of arbitrariness or discrimination,” Roman, supra, at 710. See Mahan v. Howell, 410 U. S. 315, 325-326 (1973). Third, the State must show that “the state policy urged ... to justify the divergences ... is, indeed, furthered by the plan,” id., at 326. This necessarily requires a showing that any deviations from equality are not significantly greater than is necessary to serve the State’s asserted policy; if another plan could serve that policy substantially as well while providing smaller deviations from equality, it can hardly be said that the larger deviations advance the policy. See, e. g., Kilgarlin v. Hill, 386 U. S. 120, 123-124 (1967); Mahan, supra, at 319-320, 326; Connor, supra, at 420-421. Fourth, even if the State succeeds in showing that the deviations in its plan are justified by their furtherance of a rational state policy, the court must nevertheless consider whether they are small enough to be constitutionally tolerable. “For a State’s policy urged in justification of disparity in district population, however rational, cannot constitutionally be permitted to emasculate the goal of substantial population equality.” Mahan, supra, at 326.
*853B
It takes little effort to show that Wyoming’s 1981 House of Representatives apportionment is manifestly unconstitutional under the test established by our cases, whether one considers the instance of Niobrara County alone or in combination with the large deviations present in the rest of the scheme.
It is conceded all around, of course, that appellants have shown a prima facie case of discrimination. Wyoming’s 89% maximum deviation greatly exceeds our “under 10%” threshold; indeed, so great is the inequality in this plan that even its 16% average deviation from ideal district size exceeds the threshold we have set for maximum deviations. On the other hand, one might reasonably concede that the State has met the second and third steps. Wyoming’s longstanding policy of using counties as the basic units of representation is a rational one, found by the District Court to be untainted by arbitrariness or discrimination. It appears as well that the deviations at issue could not be reduced (at least not without substantially increasing the size of the House of Representatives) consistently with Wyoming’s goals of using county lines and assuring each county at least one representative. It cannot plausibly be argued, however, that Wyoming’s plan passes the fourth test — that its deviations, even if justified by state policy, be within the constitutionally tolerable range of size.
We have warned that although maintenance of county or other political boundaries can justify small deviations, it cannot be allowed to negate the fundamental principle of one person, one vote. E. g., Connor, supra, at 419. Likewise, we have recognized that it may not always be feasible, within constitutional constraints, to guarantee each county or subdivision a representative of its own. “Carried too far, a scheme of giving at least one seat in one house to each political subdivision (for example, to each county) could easily result, in many States, in a total subversion of the equal-*854population principle in that legislative body.” Reynolds, 377 U. S., at 581 (footnote omitted); see Mahan, supra, at 349, n. 11 (Brennan, J., concurring in part and dissenting in part). And we have unambiguously rejected reliance on the very factor the State urges as the reason for its plan, stating that sparseness of population, far from excusing deviations from equality, actually increases the need for equality among districts:
“[Sjparse population is not a legitimate basis for a departure from the goal of equality. A State with a sparse population may face problems different from those faced by one with a concentrated population, but that, without more, does not permit a substantial deviation from the average. Indeed, in a State with a small population, each individual vote may be more important to the result of an election than in a highly populated State. Thus, particular emphasis should be placed on establishing districts with as exact population equality as possible.” Chapman v. Meier, 420 U. S. 1, 24-25 (1975) (emphasis added).
Accord, Connor, supra, at 418-419, n. 18; see Reynolds, supra, at 580.
As the Court implicitly acknowledges, ante, at 843, Nio-brara County’s overrepresentation — 60% compared to the ideal district size — cannot be considered “the kind of ‘minor’ variatio[n] which Reynolds v. Sims indicated might be justified by local policies counseling the maintenance of established political subdivisions in apportionment plans.” Kilgarlin, 386 U. S., at 123. In Kilgarlin, we expressed strong doubt that the 26% maximum deviation there could ever be permitted, ibid. In Mahan, we warned that a 16.4% maximum deviation, even though fully justified by state policy, “may well approach tolerable limits.” 410 U. S., at 329. See also Abate v. Mundt, 403 U. S. 182, 187 (1971). Here, by contrast, Niobrara County voters are given more than two and a half times the voting strength of the average Wyoming voter, *855and more than triple the voting strength of voters in some counties.2 “[I]f a State should provide that the votes of citizens in one part of the State should be given two times, or five times, or 10 times the weight of votes of citizens in another part of the State, it could hardly be contended that the right to vote of those residing in the disfavored areas had not been effectively diluted.” Reynolds, supra, at 562. The creation of this district represents not a deviation from the principle of population equality, but an absolute disregard of it. Niobrara County, alone in the State, has been allocated a seat “on a basis wholly unrelated to population.” WMCA, Inc. v. Lomenzo, 377 U. S. 633, 645 (1964). This hardly constitutes “a faithful adherence to a plan of population-based representation.” Roman, 377 U. S., at 710.
If the rest of the State is considered as well, the picture becomes even worse. The scheme’s treatment of Niobrara County is not a single, isolated abuse, but merely the worst of many objectionable features. Of Wyoming’s 23 counties, only 9 are within as much as 10% of population proportionality. The populations per representative of Sublette and Crook Counties are, respectively, 38% and 28% below the statewide average; those of Washakie and Teton Counties are 29% and 28%, respectively, above that figure. The average deviation from ideal district size is 16%. The figures could be spun out further, but it is unnecessary. It is not surprising, then, that the Court makes no effort to uphold the plan as a whole. On the contrary, at least two Members of the majority express their “gravest doubts that a statewide legislative plan with an 89% maximum deviation could survive *856constitutional scrutiny despite the presence of the State’s strong interest in preserving county boundaries.” Ante, at 850 (O’Connor, J., joined by Stevens, J., concurring).
C
The Court attempts to escape these stark facts through two lines of reasoning, each relying on an unspoken legal premise. Neither withstands examination.
First, the Court apparently assumes that the only aspect of unequal representation that matters is the degree of vote dilution suffered by any one individual voter. See ante, at 847. The Court is mistaken. Severe dilution of the votes of a relatively small number of voters is perhaps the most disturbing result that may attend invalid apportionments, because those unfortunate victims may be virtually disfranchised. It is not the sole evil to be combated, however. It is equally illegal to enact a scheme under which a small group is greatly overrepresented, at the expense of all other voters in the State. Such a “rotten borough”3 plan does tend to yield small figures supposedly measuring the harm to single individuals, as the Court’s opinion illustrates; but that analysis overlooks the fact that very large numbers of persons are adversely affected.4 It is the principle of equal representation, as well as the votes of individual plaintiffs, that a State may not dilute. Reynolds, supra, at 578. Just as the Equal Protection Clause does not permit a small class of voters to be deprived of fair and equal voting power, so does it forbid the elevation of a small class of “supervoters” granted an extraordinarily powerful franchise. We would not permit Wyoming, in its legislative elections, to grant a double- or triple-counted vote to 2,924 voters because they were named Jones, or because they were licensed to practice law — even though such an enactment would, by the Court’s reasoning, have *857only a de minimis effect on the rights of the rest of Wyoming’s voters. Why, then, is it permissible to create such an exalted class based on location of residence?
The Court relies more directly on its unspoken assumption that we may judge the constitutionality of Niobrara County’s representation by first severing that feature from the rest of the scheme, and then weighing it only by its incremental effect in increasing the degree of inequality present in the system as a whole.
“Appellants deliberately have limited their challenge to the alleged dilution of their voting power resulting from the one representative given to Niobrara County. The issue therefore is not whether a 16% average deviation and an 89% maximum deviation, considering the state apportionment plan as a whole, are constitutionally permissible. Rather, the issue is whether Wyoming’s policy of preserving county boundaries justifies the additional deviations from population equality resulting from the provision of representation to Niobrara County.” Ante, at 846 (footnotes omitted).
The first leg of this logic — that the Niobrara problem is legally severable from the rest of the plan — is contradicted by our prior decisions. The second leg — that we should examine only the marginal unequalizing effect — leads to exceptionally perverse results.
We confronted an analogous situation in Maryland Committee for Fair Representation v. Tawes, 377 U. S. 656 (1964). The State argued in Tawes that since the plaintiffs had allegedly conceded that one house of the Maryland Legislature was constitutionally apportioned, and the courts below had passed only on the apportionment of the other house, this Court was required to limit its consideration to the apportionment of the challenged house. We flatly rejected the argument:
“Regardless of possible concessions made by the parties and the scope of the consideration of the courts *858below, in reviewing a state legislative apportionment case this Court must of necessity consider the challenged scheme as a whole in determining whether the particular State’s apportionment plan, in its entirety, meets federal constitutional requisites. It is simply impossible to decide upon the validity of the apportionment of one house of a bicameral legislature in the abstract, without also evaluating the actual scheme of representation employed with respect to the other house. Rather, the proper, and indeed indispensable, subject for judicial focus in a legislative apportionment controversy is the overall representation accorded to the State’s voters, in both houses of a bicameral state legislature. We therefore reject [the State’s] contention that the Court is precluded from considering the validity of the apportionment of the Maryland House of Delegates.” Id., at 673.
Accord, Lucas v. Colorado General Assembly, 377 U. S. 713, 735, n. 27 (1964).5
Although we have not invariably adhered to this rule with regard to the two houses of a legislature, the concerns that led us in Tawes to examine both houses, despite the scope of the plaintiffs’ complaint, forbid us to consider the allocation of one seat without also examining the remainder of Wyoming’s apportionment of its House of Representatives. A plan with only a single deviation — a good deal smaller than this one, *859and necessary to carry out a rational state policy — might well be tolerated, even though in the same situation a greater number of substantial deviations would be unacceptable as too much of a departure from the goal of equality. See Lucas, supra, at 735, n. 27. Where that greater number of deviations is present, as in this case, common sense as well as Tawes and Lucas require us to consider the plan as a whole. The inequality created by Niobrara County’s representation— a 23% increase in the maximum deviation from equality — is necessarily cumulative with the inequality imposed in the rest of the system. It is playing artificial tricks to assert that the fairness of the allocation of one seat in a legislative body can or should be considered as though it had no connection to the other seats, or to the fairness of their allocation. Indeed, the Court’s own method contradicts its suggestion that the Niobrara problem is severable. The Court is fully willing to consider the system’s other inequalities in this case, and even to give them controlling weight — only it wishes to consider those inequalities as weighing in favor of the plan. See infra, this page and 860. I agree with the Court that we may not consider Niobrara County in a vacuum; it seems to me, however, that the existence of numerous instances of inequality ought to be considered an undesirable feature in an apportionment plan, not a saving one. Only by examining the plan “in its totality,” Lucas, supra, at 735, n. 27, may we judge whether the allocation of any seat in the House is constitutional. This Court is not bound by a referendum of the League of Women Voters. See ante, at 846, n. 8.
Here, Wyoming’s error in granting Niobrara County voters a vote worth double or triple the votes of other Wyoming voters is compounded by the impermissibly large disparities in voting power existing in the rest of the apportionment plan. Supra, at 855. Yet, astonishingly, the Court manages to turn that damning fact to the State’s favor:
“The allocation of a representative to a particular political subdivision still may violate the Equal Protection Clause if it greatly exceeds the population variations ex*860isting in the rest of the State and if the State provides no legitimate justifications for the creation of that seat. Here, however, considerable population variations will remain even if Niobrara County's representative is eliminated. . . . These statistics make clear that the grant of a representative to Niobrara County is not a significant cause of the population deviations that exist in Wyoming.” Ante, at 847.
Under this reasoning, the further Wyoming’s apportionment plan departs from substantial equality, the more likely it is to withstand constitutional attack. It is senseless to create a rule whereby a single instance of gross inequality is unconstitutional if it occurs in a plan otherwise letter-perfect, but constitutional if it occurs in a plan that, even without that feature, flagrantly violates the Constitution. That, however, is precisely what the Court does today.6
*861D
Justice O’Connor, joined by Justice Stevens, states that she has “the gravest doubts that a statewide legislative plan with an 89% maximum deviation could survive constitutional scrutiny. . . Ante, at 850 (concurring opinion). But the Court today holds that just such a plan does survive constitutional scrutiny. I dissent.
As the Court notes, of course, we have been substantially more demanding with respect to apportionment of federal congressional districts. Mahan v. Howell, 410 U. S. 315, 320-325 (1973). See generally Karcher v. Daggett, ante, p. 725; White v. Weiser, 412 U. S. 783 (1973); Kirkpatrick v. Preisler, 394 U. S. 526 (1969).
The ideal district size — statewide population divided by number of seats — is 7,337; Niobrara County’s population is 2,924. Thus, the average representative represents 2.59 times as many constituents as Niobrara County’s representative. Similarly, the populations of Washakie and Teton Counties are, respectively, 3.25 and 3.19 times as large as the population of Niobrara County, yet all three counties are given one representative each. 1 App. Exhibits 19-20.
See generally Reynolds v. Sims, 377 U. S. 533, 567-568, n. 44 (1964); Baker v. Carr, 369 U. S. 186, 302-307 (1962) (Frankfurter, J., dissenting).
Cf. Swann v. Adams, 385 U. S. 440, 443 (1967).
“[In] Maryland Committee for Fair Representation v. Tawes, ... we discussed the need for considering the apportionment of seats in both houses of a bicameral state legislature in evaluating the constitutionality of a state legislative apportionment scheme, regardless of what matters were raised by the parties and decided by the court below. Consistent with this approach, in determining whether a good faith effort to establish districts substantially equal in population has been made, a court must necessarily consider a State’s legislative apportionment scheme as a whole. Only after evaluation of an apportionment plan in its totality can a court determine whether there has been sufficient compliance with the requisites of the Equal Protection Clause.” 377 U. S., at 785, n. 27 (emphasis added). See also Burns v. Richardson, 384 U. S. 73, 83 (1966).
This case also presents an issue as to what relief should be accorded. At an absolute minimum, the District Court should have granted the relief requested by appellants — the combination of Niobrara and Goshen Counties into one district, as provided by the Wyoming Legislature in case its first plan was found unconstitutional. See ante, at 840. That would have yielded a combined district of virtually perfect size, and would have reduced the plan’s maximum deviation from 89% to 66%. This improvement alone — 23%—is larger than any maximum deviation we have ever approved, with or without justification. See supra, at 854.
In my view, however, the District Court should have required Wyoming to devise an apportionment plan constitutional in its entirety. In Whitcomb v. Chavis, 403 U. S. 124 (1971), the plaintiffs’ complaint attacked Indiana’s apportionment statute only as to one county. Id., at 137. We reversed the District Court’s judgment that that county was unconstitutionally apportioned. Nevertheless, we expressly approved the District Court’s decision to expand the relief granted to include reapportionment of the entire State. “After determining that Marion County required reapportionment, the court concluded that ‘it becomes clear beyond question that the evidence adduced in this case and the additional apportionment requirements set forth by the Supreme Court call for a redistricting of the entire state as to both houses of the General *861Assembly.’” Id., at 161 (plurality opinion), quoting 305 F. Supp. 1364, 1391 (SD Ind. 1969); see 403 U. S., at 172-173, 179-180 (Douglas, J., concurring in result in part). See also supra, at 857-859, and n. 5; Fed. Rule Civ. Proc. 54(c).