Mr. Justice Stewart

delivered the opinion of the Court.

New York law provides that a new county charter will go into effect only if it is approved in a referendum election by separate majorities of the voters who live in the cities within the county, and of those who live outside the cities. A three-judge Federal District Court held that these requirements violate the Equal Protection Clause of the Fourteenth Amendment. We noted probable jurisdiction of this direct appeal from the District Court’s judgment under 28 U. S. C. § 1253. 426 U. S. 918.

County government in New York has traditionally taken the form of a single-branch legislature, exercising general governmental powers. General governmental powers are also exercised by the county’s constituent cities, villages, and towns. The allocation of powers among these subdivisions can be changed, and a new form of county government adopted, pursuant to referendum procedures specified in Art. IX of the New York Constitution 1 and implemented by § 33 *261 of the Municipal Home Rule Law. 2 Under those procedures a county board of supervisors may submit a proposed charter to the voters for approval. If a majority of the voting city *262 dwellers and a majority of the voting noncity dwellers both approve, the charter is adopted. 3

In November 1972, a proposed charter for the county of Niagara was put to referendum. The charter created the new offices of County Executive and County Comptroller, and continued the county’s existing power to establish tax rates, equalize assessments, issue bonds, maintain roads, and administer health and public welfare services. No explicit provision for redistribution of governmental powers from the cities or towns to the county government was made. The city voters approved the charter by a vote of 18,220 to 14,914. The noncity voters disapproved the charter by a vote of 11,594 to 10,665. 4 A majority of those voting in the entire county thus favored the charter. 5

*263 The appellees, a group of Niagara County voters, filed suit pursuant to 42 U. S. C. § 1983 in the United States District Court for the Western District of New York, seeking a declaration that the New York constitutional and statutory provisions governing adoption of the charter form of county government are unconstitutional, and an order directing the appropriate New York officials to file the Niagara County charter as a duly enacted local law. A three-judge court was convened. Before its decision was announced, however, another new charter was put to referendum in Niagara County in November 1974. Again a majority of the city dwellers who voted approved the charter, a majority of the noncity voters disapproved it, and an aggregate majority of all those in the county who voted approved it. 6 The District Court subsequently found the concurrent-majority requirements of the New York Constitution and the New York Municipal Home Rule Law violative of the Equal Protection Clause of the Fourteenth Amendment, and ordered implementation of the 1972 Charter. 386 F. Supp. 1. 7 On appeal this Court vacated that judgment and remanded the cause “for reconsideration in light of the provisions of [the] new charter adopted by Niagara County in 1974.” 423 U. S. 808. In subsequent proceedings on remand, the District Court found that there was “no substantial difference between the *264 two Charters” and that the 1974 County Charter had superseded the 1972 Charter. 8 Pursuant to its previous constitutional adjudication, the court decreed that the 1974 Charter “is in full force and effect as the instrument defining the form of local government for Niagara County.” 9

The impact of the Equal Protection Clause on the exercise of the electoral franchise under state law is hardly a novel concern of the federal judiciary. It was made clear more than 15 years ago in Baker v. Carr, 369 U. S. 186, that the subject is a justiciable one, and ever since the seminal case of Reynolds v. Sims, 377 U. S. 533, it has been established that the Equal Protection Clause cannot tolerate the disparity in individual voting strength that results when elected officials represent districts of unequal population, since “the funda *265 mental principle of representative government in this country is one of equal representation for equal numbers of people, without regard to race, sex, economic status, or place of residence within a State.” Id., at 560-561. 10

In the case before us the District Court, though recognizing that “the precise issue here presented appears to be one of first impression,” concluded that the rule of Reynolds v. Sims, controlled its resolution. “Reasoning by analogy,” the court held, in short, that the dual-majority requirement of New York law “is unconstitutional because it violates the one man, one vote principle.” 386 F. Supp., at 7. In assessing the correctness of the District Court’s judgment it is thus appropriate to begin by recalling the basic rationale of the decisions of this Court in which that principle was first developed and applied.

The rationale is, at bottom, so simple as to be almost self-evident. Beginning with Reynolds v. Sims, supra, cases in which the principle emerged involved challenges to state legislative apportionment systems that gave “the same number of representatives to unequal numbers of constituents.” 377 U. S., at 563. The Court concluded that in voting for their legislators, all citizens have an equal interest in representative democracy, and that the concept of equal protection therefore requires that their votes be given equal weight. 11 See, e. g., Lucas v. Colorado Gen. Assembly, 377 U. S. 713; Fortson v. Dorsey, 379 U. S. 433; Burns v. Richardson, 384 *266 U. S. 73; Swann v. Adams, 385 U. S. 440; Kilgarlin v. Hill, 386 U. S. 120; Whitcomb v. Chavis, 403 U. S. 124; Gaffney v. Cummings, 412 U. S. 735.

The equal protection principles applicable in gauging the fairness of an election involving the choice of legislative representatives are of limited relevance, however, in analyzing the propriety of recognizing distinctive voter interests in a “single-shot” referendum. In a referendum, the expression of voter will is direct, and there is no need to assure that the voters’ views will be adequately represented through their representatives in the legislature. The policy impact of a referendum is also different in kind from the impact of choosing representatives—instead of sending legislators off to the state capitol to vote on a multitude of issues, the referendum puts one discrete issue to the voters. That issue is capable, at least, of being analyzed to determine whether its adoption or rejection will have a disproportionate impact on an identifiable group of voters. If it is found to have such a disproportionate impact, the question then is whether a State can recognize that impact either by limiting the franchise to those voters specially affected or by giving their votes a special weight. This question has been confronted by the Court in two types of cases: those dealing with elections involving “special-interest” governmental bodies of limited jurisdiction, and those dealing with bond referenda.

The Court has held that the electorate of a special-purpose unit of government, such as a water storage district, may be apportioned to give greater influence to the constituent groups found to be most affected by the governmental unit’s functions. Salyer Land Co. v. Tulare Water Dist., 410 U. S. 719. But the classification of voters into “interested” and “noninterested” groups must still be reasonably precise, as Kramer v. Union School Dist., 395 U. S. 621, demonstrates. The Court assumed in that case that the voting constituency in school district elections could be limited to those “primarily inter *267 ested in school affairs,” id., at 632, but concluded that the State’s classification of voters on the asserted basis of that interest was so imprecise that the exclusion of otherwise qualified voters was impermissible. 12

In the bond referenda cases, the local government had either limited the electoral franchise to property owners, or weighted property owners’ votes more heavily than those of nonproperty owners by using a “dual box” separate-majority approval system quite similar to the one at issue in the present case. Cipriano v. City of Houma, 395 U. S. 701; Phoenix v. Kolodziejski, 399 U. S. 204; Hill v. Stone, 421 U. S. 289.

In the Cipriano case, involving revenue bonds, it was apparent that all voters had an identity of interest in passage of the bond issue, and limitation of the electoral franchise to “property taxpayers” was, plainly, invidiously discriminatory. The other two cases, however, involved general obligation bonds. There, as in Salyer and Kramer, the validity of the classification depended upon whether the group interests were sufficiently different to justify total or partial withholding of the electoral franchise from one of them. In support of the classifications, it was argued that property owners have a more substantial stake in the adoption of obligation bonds than do nonproperty owners, because the taxes of the former directly and substantially fund the bond obligation. The Court rejected that argument for limiting the electoral franchise, however, noting that nonproperty owners also share in the tax burden when the tax on rental property or commercial businesses is passed on in the form of higher prices. Although the interests of the two groups are concededly not *268 identical, the Court held that they are sufficiently similar to prevent a state government from distinguishing between them by artificially narrowing or weighting the electoral franchise in favor of the property taxpayers. 13

These decisions do not resolve the issues in the present case. Taken together, however, they can be said to focus attention on two inquiries: whether there is a genuine difference in the relevant interests of the groups that the state electoral classification has created; and, if so, whether any resulting enhancement of minority voting strength nonetheless amounts to invidious discrimination in violation of the Equal Protection Clause.

III

The argument that the provisions of New York law in question here are unconstitutional rests primarily on the premise that all voters in a New York county have identical interests in the adoption or rejection of a new charter, and that any distinction, therefore, between voters drawn on the basis of residence and working to the detriment of an identifiable class is an invidious discrimination. If the major premise were demonstrably correct—if it were clear that all voters in Niagara County have substantially identical interests in the adoption of a new county charter, regardless of where they reside within the county—the District Court’s judgment would have to be affirmed under our prior cases. Cipriano v. City of Houma, supra. That major premise, however, simply cannot be accepted. To the contrary, it appears that the challenged provisions of New York law rest on the State’s identification of the distinctive interests of the residents of *269 the cities and towns within a county rather than their interests as residents of the county as a homogeneous unit. This identification is based in the realities of the distribution of governmental powers in New York, and is consistent with our cases that recognize both the wide discretion the States have in forming and allocating governmental tasks to local subdivisions, and the discrete interests that such local governmental units may have qua units. Reynolds v. Sims, 377 U. S., at 580; Abate v. Mundt, 403 U. S. 182; Mahan v. Howell, 410 U. S. 315.

General-purpose local government in New York is entrusted to four different units: counties, cities, towns, and villages. The State is divided into 62 counties; each of the 57 counties outside of New York City is divided into towns, or towns and one or more cities. Villages, once formed, are still part of the towns in which they are located. The New York Legislature has conferred home rule and general governmental powers on all of these subdivisions, and their governmental activities may on occasion substantially overlap. 14 The cities often perform functions within their jurisdiction that the county may perform for noncity residents; similarly villages perform some functions for their residents that the town provides for the rest of the town’s inhabitants. Historically towns provided their areas with major social services that more recently have been transferred to counties; towns exercise more regulatory power than counties; and both towns and counties can create special taxing and improvement districts to administer services. See 13 New York Temporary State Commission on the Constitutional Convention, Local Government 20 (1967).

*270 Acting within a fairly loose state apportionment of political power, the relative energy and organization of these various subdivisions will often determine which one of them in a given area carries out the major tasks of local government. Since the cities have the greatest autonomy within this scheme, changes serving to strengthen the county structure may have the most immediate impact on the functions of the towns as deliverers of government services. Id., at 19. 15

The provisions of New York law here in question clearly contemplate that a new or amended county charter will frequently operate to transfer “functions or duties” from the towns or cities to the county, or even to “abolish one or more offices, departments, agencies or units of government.” 16 Although the 1974 Charter does not explicitly transfer governmental functions or duties from the towns to Niagara County, the executive-legislative form of government it provides would significantly enhance the county’s organizational and service delivery capacity, for the purpose of “greater efficiency and responsibility in county government.” Niagara County Charter, 1972. The creation of the offices of County Executive and Commissioner of Finance clearly reflects this purpose. Such anticipated organizational changes, no less than explicit transfers of functions, could effectively shift any pre-existing balance of power between town and county governments toward county predominance. 17 In terms of efficient delivery *271 of government services, such a shift might be all to the good, but it may still be viewed as carrying a cost quite different for town voters and their existing town governments from that incurred by city voters and their existing city governments.

The ultimate question then is whether, given the differing interests of city and noncity voters in the adoption of a new county charter in New York, those differences are sufficient under the Equal Protection Clause to justify the classifications made by New York law. Phoenix v. Kolodziejski, 399 U. S. 204; Salyer Land Co. v. Tulare Water Dist., 410 U. S. 719; Hill v. Stone, 421 U. S. 289. If that question were posed in the context of annexation proceedings, the fact that the residents of the annexing city and the residents of the area to be annexed formed sufficiently different constituencies with sufficiently different interests could be readily perceived. The fact of impending union alone would not so merge them into one community of interest as constitutionally to require that their votes be aggregated in any referendum to approve annexation. Cf. Hunter v. Pittsburgh, 207 U. S. 161. Similarly a proposal that several school districts join to form a consolidated unit could surely be subject to voter approval in each constituent school district.

Yet in terms of recognizing constituencies with separate and potentially opposing interests, the structural decision to annex or consolidate is similar in impact to the decision to restructure county government in New York. In each case, separate voter approval requirements are based on *272 the perception that the real and long-term impact of a restructuring of local government is felt quite differently by the different county constituent units that in a sense compete to provide similar governmental services. Voters in these constituent units are directly and differentially affected by the restructuring of county government, which may make the provider of public services more remote and less subject to the voters’ individual influence.

The provisions of New York law here in question no more than recognize the realities of these substantially differing electoral interests. 18 Granting to these provisions the presumption of constitutionality to which every duly enacted state and federal law is entitled, 19 we are unable to *273 conclude that they violate the Equal Protection Clause of the Fourteenth Amendment.

For the reasons stated in this opinion the judgment is reversed.

It is so ordered.

The Chief Justice concurs in the judgment.

Notes

Article IX, § 1 (h)(1), of the New York Constitution (McKinney 1969) provides in pertinent part:

“§ 1. Bill of rights for local governments

“Effective local self-government and intergovernmental cooperation are purposes of the people of the state. In furtherance thereof, local govern *261 ments shall have the following rights, powers, privileges and immunities in addition to those granted by other provisions of this constitution:

“(h)(1) Counties, other than those wholly included within a city, shall be empowered . . . to adopt, amend or repeal alternative forms of county government . . . . Any such form of government or any amendment thereof . . . may transfer one or more functions or duties of the county or of the cities, towns, villages, districts or other units of government wholly contained in such county to each other . . . or may abolish one or more offices, departments, agencies or units of government provided, however, that no such form or amendment . . . shall become effective unless approved on a referendum by a majority of the votes cast thereon in the area of the county outside of cities, and in the cities of the county, if any, considered as one unit. Where an alternative form of county government or any amendment thereof . . . provides for the transfer of any function or duty to or from any village or the abolition of any office, department, agency or unit of government of a village wholly contained in such county, such form or amendment shall not become effective unless it shall also be approved on the referendum by a majority of the votes cast thereon in all the villages so affected considered as one unit.”

Section 33 (7) of the Municipal Home Rule Law of New York (McKinney Supp. 1976-1977) provides:

“§ 33. Power to adopt, amend and repeal county charters

“7. A charter law

“(a) providing a county charter, or

“(b) proposing an amendment or repeal of one or more provisions thereof which would have the effect of transferring a function or duty of the county, or of a city, town, village, district or other unit of local government wholly contained in the county, shall conform to and be subject to consideration by the board of supervisors in accordance with the provisions of this chapter generally applicable to the form of and action on proposed local laws by the board of supervisors. If a county charter, or a charter law as described in this subdivision, is adopted by the board of supervisors, it shall not become operative unless and until it is approved at a general election or at a special election, held in the county by receiving a majority of the total votes cast thereon (a) in the area of the county outside of *262 cities and (b) in the area of the cities of the county, if any, considered as one unit, and if it provides for the transfer of any function or duty to or from any village or for the abolition of any office, department, agency or unit of government of a village wholly contained in the county, it shall not take effect unless it shall also receive a majority of all the votes cast thereon in all the villages so affected considered as one unit. Such a county charter or charter law shall provide for its submission to the electors of the county at the next general election or at a special election, occurring not less than sixty days after the adoption thereof by the board of supervisors. Such a county charter or charter law may provide for the separate submission to the electors at such election of one or more variations of the provisions of such county charter. Any such variation may include, but shall not be limited to, proposed transfers of functions of local government to other units of local government or a class or classes thereof.”

When the proposed charter provides for the transfer of duties or functions of villages within the county, the voters residing in the villages must also approve it by a separate majority. See nn. 1 and 2, supra. That requirement is not directly involved in the present litigation.

Village residents are also residents of the towns in which they are located, are subject to their governments, and are included in the noncity vote.

A total of 55,393 votes was cast; in the aggregate 28,885 voters favored the charter and 26,508 opposed it. The population of Niagara County was approximately 236,000.

A total of 36,808 votes was cast in the referendum election. The city dwellers who voted favored the charter by a margin of 11,305 to 9,222; the noncity dwellers voted 8,222 to 8,059 against it. In the county as a whole, 19,364 voted for the charter, and 17,444 against it.

Niagara County itself had earlier brought an action in the same court to enforce the 1972 Charter as the law of Niagara County. County of Niagara v. New York, Civ. No. 1972-656 (WDNY, Apr. 13, 1973). The District Court had dismissed the complaint on the ground that it presented no substantial federal question, and no appeal had been taken. The appellants interposed a res judicata defense based upon that judgment. The District Court properly rejected that defense upon the ground that the plaintiffs had not been parties to the earlier suit and were not in privity with the county of Niagara, which had brought it. 386 F. Supp., at 5.

The District Court also enjoined pending state proceedings brought by the appellants to challenge the certification and enforcement of the 1974 Charter. The appellants now argue that the District Court should have deferred to the jurisdiction of the state court. Even assuming that Younger v. Harris, 401 U. S. 37, principles are fully applicable in the civil context, however, the original action challenging the dual-majority provisions of New York law had been brought in the federal court well before the appellants filed their state-court suit, and principles of comity and federalism do not require that a federal court abandon jurisdiction it has properly acquired simply because a similar suit is later filed in a state court.

The District Court’s opinion on remand is unreported. The appellants argue that the relief originally sought in this suit was limited to the 1972 Charter, and that the case is now moot because the 1974 Charter has superseded it. Pursuant to our mandate on remand, however, the District. Court gave full consideration to the 1974 Charter, concluded that it was substantially the same as the 1972 Charter, and amended its judgment to recognize the validity of the 1974 Charter. Although the better practice might have been to require amendment of the complaint so as formally to seek the relief ultimately granted, the appellants were not prejudiced by the procedure adopted by the District Court, and we consider the validity of the 1974 Charter as the issue before us.

In the application of this settled rule, the Court has, of course, frequently been divided with respect to what deviations from numerical exactitude are constitutionally permissible. See, e. g., Abate v. Mundt, 403 U. S. 182; Mahan v. Howell, 410 U. S. 315; Gaffney v. Cummings, 412 U. S. 735; White v. Regester, 412 U. S. 755.

The Court has applied the principle of “one person, one vote” to representative elections at the local as well as state level. See, e. g., Avery v. Midland County, 390 U. S. 474; Hadley v. Junior College Dist., 397 U. S. 50.

See also Carrington v. Rash, 380 U. S. 89, holding that although the requirement that a citizen be a resident of Texas was clearly an acceptable means of defining relevant constituencies, Texas could not exclude all servicemen from voting on the conclusive presumption that their necessarily transient status precluded them from being bona fide residents.

We have held, however, that a referendum voting scheme that can be characterized in mathematical terms as giving disproportionate power to a minority does not violate the Equal Protection Clause, there being no discrimination against an identifiable class. Gordon v. Lance, 403 U. S. 1. Cf. Hunter v. Erickson, 393 U. S. 385.

See generally N. Y. Mun. Home Rule Law (McKinney 1969 and Supp. 1976-1977); 13 New York Temporary State Commission on the Constitutional Convention, Local Government (1967); Moore, Early History of Town Government in New York State, in N. Y. Town Law vii (McKinney 1965).

The Court has previously had occasion to recognize the historic functional interdependence of county and town subdivisions in New York, and has allowed considerable deviation from the basic rule of Reynolds v. Sims, 377 U. S. 533, on that ground. Abate v. Mundt, 403 U. S. 182.

See nn. 1 and 2, supra.

Appellee Citizens for Community Action at the Local Level argue that appellee Niagara County’s failure to controvert the allegation in the original complaint that “no specific group of voters, residents or other persons were primarily affected or interested as compared to another group” establishes that the city and town voters here had identical interests *271 in the charter’s adoption. As a factual matter, this assertion appears contradicted by the later intervention of the town of Lockport to protect its voters’ special interests in the issue. In any case, the question of the constitutionality of Art. IX of the New York Constitution and § 33 of the New York Municipal Home Rule Law turns, not on the perceptions of voters in a particular county, but on whether the State might legitimately view their interests as sufficiently different to justify a distinction between city and town voters.

There is no indication that the classifications created by New York law work to favor city voter over town voter, or town voter over city voter. In some New York counties, city voters outnumber town voters; in other counties, the reverse is true. We are advised that of charters proposed in 14 counties, one failed to obtain majority approval of the city voters; two (including Niagara County) failed to obtain majority approval of noncity voters; eight failed to obtain a majority vote in either the towns or the cities; and three were approved by both city and town voters.

The constitutional and statutory provisions in this case also do not appear to be the sustained product of either an entrenched minority or a willful majority. Instead they have been subject historically to fairly frequent revision. The constitutional amendment requiring city and town voter approval for changes in county government was first adopted by referendum in 1935. N. Y. Const., Art. III, § 26 (1935). In 1938, the Constitution was amended to provide for approval by the voters in the county as a whole, unless the proposal provided for transfers of functions to or from cities, towns, or villages in the county. N. Y. Const., Art. IX (1938). In 1958, Art. IX was again amended to provide that any change in county government had to be approved by a majority of the noncity voters, voting as a group, and by a majority of the city voters, voting as a group. The existing Art. IX, adopted in 1963, contains the same general provision as the 1958 amendment.

See Thayer, The Origin and Scope of the American Doctrine of Constitutional Law, 7 Harv. L. Rev. 129 (1893).