with whom Justice Stevens joins, concurring in part and dissenting in part.
This case presents the question whether a state legislature violates the Equal Protection Clause by adopting a redistricting plan designed solely to preserve the power of the dominant political party, when the plan follows the doctrine of “one person, one vote” but ignores all other neutral factors relevant to the fairness of redistricting.1
In answering this question, the plurality expresses the view, with which I agree, that a partisan political gerrymander violates the Equal Protection Clause only on proof of “both intentional discrimination against an identifiable political group and an actual discriminatory effect on that group.” Ante, at 127. The plurality acknowledges that the record in this case supports a finding that the challenged redistricting plan was adopted for the purpose of discriminating against Democratic voters. Ibid. The plurality argues, however, *162that appellees failed to establish that their voting strength was diluted statewide despite uncontradicted proof that certain key districts were grotesquely gerrymandered to enhance the election prospects of Republican candidates. This argument appears to rest solely on the ground that the legislature accomplished its gerrymander consistent with “one person, one vote,” in the sense that the legislature designed voting districts of approximately equal population and erected no direct barriers to Democratic voters’ exercise of the franchise. Since the essence of a gerrymandering claim is that the members of a political party as a group have been denied their right to “fair and effective representation,” Reynolds v. Sims, 377 U. S. 533, 565 (1964), I believe that the claim cannot be tested solely by reference to “one person, one vote.” Rather, a number of other relevant neutral factors must be considered. Because the plurality ignores such factors and fails to enunciate standards by which to determine whether a legislature has enacted an unconstitutional gerrymander, I dissent.
I
The facts are exhaustively described in the District Court’s opinion and may be briefly restated here. In 1981, the Republican Party controlled both houses of the Indiana General Assembly, and its candidate held the Governor’s seat. Pursuant to the requirements of the State Constitution, the General Assembly undertook legislative redistricting based on 1980 census data. A Conference Committee, all of whose members were Republicans, was assigned the task of drawing district maps with the assistance of a private computer firm. The information fed into the computer primarily concerned the political complexion of the State’s precincts. The redistricting process was conducted in secret. Democratic legislators were not afforded any participation in designing the district maps that were adopted. There were no hearings where members of the public were invited to express *163their views. The Republican Committee revealed its proposed redistrieting plan two days before the end of the legislative session, and the Democrats hurriedly presented an alternative plan. On the last day of the session, the Republican plan was adopted by party line vote in both Houses of the General Assembly. The Governor signed the plan into law.
In 1982 and 1984, elections were held under the new redistricting plan. Prior to the 1982 election, this lawsuit was commenced by appellees, a group of Indiana Democrats who claimed that the plan constitutes a partisan political gerrymander designed to disenfranchise Democratic voters in violation of the Equal Protection Clause of the Fourteenth Amendment.2 Since trial was completed after the 1982 election, appellees relied in part on the disparity between votes cast for Democratic legislative candidates in that election and seats captured by Democrats. The case was heard by a three-judge panel in the District Court for the Southern District of Indiana. The District Court, over the dissent of Judge Pell, made extensive findings of fact and determined that appellees had established an unconstitutional partisan gerrymander. 603 F. Supp. 1479 (SD Ind. 1984). The *164Court today reverses the District Court, without concluding that any of its findings was clearly erroneous.
I — I hH
A
Gerrymandering is “the deliberate and arbitrary distortion of district boundaries and populations for partisan or personal political purposes.” Kirkpatrick v. Preisler, 394 U. S. 526, 538 (1969) (Fortas, J., concurring).3 As Justice Stevens correctly observed, gerrymandering violates the Equal Protection Clause only when the redistricting plan serves “no purpose other than to favor one segment — whether racial, ethnic, religious, economic, or political — that may occupy a position of strength at a particular time, or to disadvantage a politically weak segment of the community.” Karcher v. Daggett, 462 U. S. 725, 748 (1983) (concurring opinion).
The term “gerrymandering,” however, is also used loosely to describe the common practice of the party in power to choose the redistricting plan that gives it an advantage at the polls. An intent to discriminate in this sense may be present whenever redistricting occurs. See Gaffney v. Cummings, 412 U. S. 735, 753 (1973); Cousins v. City Council of Chicago, 466 F. 2d 830, 847 (CA7) (Stevens, J., dissenting), cert. denied, 409 U. S. 893 (1972). Moreover, since legislative *165bodies rarely reflect accurately the popular voting strength of the principal political parties, the effect of any particular redistricting may be perceived as unfair. See 412 U. S., at 752-754. Consequently, only a sensitive and searching inquiry can distinguish gerrymandering in the “loose” sense from gerrymandering that amounts to unconstitutional discrimination. Because it is difficult to develop and apply standards that will identify the unconstitutional gerrymander, courts may seek to avoid their responsibility to enforce the Equal Protection Clause by finding that a claim of gerrymandering is nonjusticiable. I agree with the Court that such a course is mistaken, and that the allegations in this case raise a justiciable issue.4
Moreover, I am convinced that appropriate judicial standards can and should be developed. Justice Fortas’ definition of unconstitutional gerrymandering properly focuses on whether the boundaries of the voting districts have been distorted deliberately and arbitrarily to achieve illegitimate ends. Kirkpatrick v. Preisler, supra, at 538. Under this definition, the merits of a gerrymandering claim must be determined by reference to the configurations of the districts, the observance of political subdivision lines, and other criteria that have independent relevance to the fairness of redistricting. See Karcher v. Daggett, supra, at 755-759 (Stevens, J., concurring). In this case, the District Court examined the redistricting in light of such factors and found, among other facts, that the boundaries of a number of districts were deliberately distorted to deprive Democratic voters of an equal opportunity to participate in the State’s legislative processes. The plurality makes no reference to any of these findings of fact. It rejects the District Court’s *166ultimate conclusion with no explanation of the respects in which appellees’ proof fell short of establishing discriminatory effect. A brief review of the Court’s jurisprudence in the context of another kind of challenge to redistricting, a claim of malapportionment, demonstrates the pressing need for the Court to enunciate standards to guide legislators who redistrict and judges who determine the constitutionality of the legislative effort.
B
The Equal Protection Clause guarantees citizens that their State will govern them impartially. See Karcher v. Daggett, supra, at 748 (Stevens, J., concurring). In the context of redistricting, that guarantee is of critical importance because the franchise provides most citizens their only voice in the legislative process. Reynolds v. Sims, 377 U. S., at 561-562, 565-566. Since the contours of a voting district powerfully may affect citizens’ ability to exercise influence through their vote, district lines should be determined in accordance with neutral and legitimate criteria. When deciding where those lines will fall, the State should treat its voters as standing in the same position, regardless of their political beliefs or party affiliation. Chapman v. Meier, 420 U. S. 1, 17 (1975); Gaffney v. Cummings, supra, at 751.
The first cases in which this Court entertained equal protection challenges to redistricting involved allegations that state legislatures had refused to redesign States’ voting districts to eliminate gross population disparities among those districts. E. g., Baker v. Carr, 369 U. S. 186 (1962); Reynolds v. Sims, supra. The Court’s decision in Reynolds v. Sims illustrates two concepts that are vitally important in evaluating an equal protection challenge to redistricting. First, the Court recognized that equal protection encompasses a guarantee of equal representation, requiring a State to seek to achieve through redistricting “fair and effective representation for all citizens.” Reynolds v. Sims, supra, at *167565-566; see Gaffney v. Cummings, 412 U. S., at 748. The concept of “representation” necessarily applies to groups: groups of voters elect representatives, individual voters do not. Gross population disparities violate the mandate of equal representation by denying voters residing in heavily populated districts, as a group, the opportunity to elect the number of representatives to which their voting strength otherwise would entitle them. While population disparities do dilute the weight of individual votes, their discriminatory effect is felt only when those individual votes are combined. Thus, the fact that individual voters in heavily populated districts are free to cast their ballot has no bearing on a claim of malapportionment.
Second, at the same time that it announced the principle of “one person, one vote” to compel States to eliminate gross disparities among district populations, the Court plainly recognized that redistricting should be based on a number of neutral criteria, of which districts of equal population was only one. Reynolds v. Sims identified several of the factors that should guide a legislature engaged in redistricting. For example, the Court observed that districts should be compact and cover contiguous territory, precisely because the alternative, “[indiscriminate districting,” would be “an open invitation to partisan gerrymandering.” 377 U. S., at 578-579. Similarly, a State properly could choose to give “independent representation” to established political subdivisions. Adherence to community boundaries, the Court reasoned, would both “deter the possibilities of gerrymandering,” and allow communities to have a voice in the legislature that directly controls their local interests. Id., at 580-581. See also Mahan v. Howell, 410 U. S. 315, 325-326 (1973). Thus, Reynolds v. Sims contemplated that “one person, one vote” would be only one among several neutral factors that serve the constitutional mandate of fair and effective representa*168tion. See Gaffney v. Cummings, supra, at 748-749. It was not itself to be the only goal of redistricting.5
A standard that judges the constitutionality of a district-ing plan solely by reference to the doctrine of “one person, one vote” may cause two detrimental results.6 First, as a perceived way to avoid litigation, legislative bodies may place undue emphasis on mathematical exactitude, subordinating or ignoring entirely other criteria that bear directly on the fairness of redistricting. See Karcher v. Daggett, 462 U. S., at 753 (Stevens, J., concurring); id., at 774 (White, J., dissenting); Gaffney v. Cummings, supra, at 749. Second, as this case illustrates, and as Reynolds v. Sims anticipated, exclusive or primary reliance on “one person, one vote” can betray the constitutional promise of fair and effective representation by enabling a legislature to engage intentionally in clearly discriminatory gerrymandering. See Wells v. Rockefeller, 394 U. S. 542, 551 (1969) (Harlan, J., dissenting).
*169C
In light of the foregoing principles, I believe that the plurality’s opinion is seriously flawed in several respects. First, apparently to avoid the forceful evidence that some district lines indisputably were designed to and did discriminate against Democrats, the plurality describes appellees’ claim as alleging that “Democratic voters over the State as a. whole, not Democratic voters in particular districts, have been subjected to unconstitutional discrimination.” Ante, at 127. This characterization is not inconsistent with appellees’ proof, and the District Court’s finding, of statewide discriminatory effect resulting from “individual districting” that “exempli-f[ies] this discrimination.” Ibid. If Democratic voters in a number of critical districts are the focus of unconstitutional discrimination, as the District Court found, the effect of that discrimination will be felt over the State as a whole.
The plurality also erroneously characterizes the harm members of the losing party suffer as a group when they are deprived, through deliberate and arbitrary distortion of district boundaries, of the opportunity to elect representatives of their choosing.7 It may be, as the plurality suggests, that *170representatives will not “entirely ignore the interests” of opposition voters. Ante, at 132. But it defies political reality to suppose that members of a losing party have as much political influence over state government as do members of the victorious party. Even the most conscientious state legislators do not disregard opportunities to reward persons or groups who were active supporters in their election campaigns. Similarly, no one doubts that partisan considerations play a major role in the passage of legislation and the appointment of state officers. Not surprisingly, therefore, the District Court expressly found that “[cjontrol of the General Assembly is crucial” to members of the major political parties in Indiana. 603 F. Supp., at 1483.8 In light of those findings, I cannot accept the plurality’s apparent conclusion that loss of this “crucial” position is constitutionally insignificant as long as the losers are not “entirely ignored” by the winners.
The plurality relies almost exclusively on the “one person, one vote” standard to reject appellees’ convincing proof that the redistricting plan had a seriously discriminatory effect on their voting strength in particular districts. The plurality properly describes the claim in this case as a denial of fair and effective “representation,” ante, at 124, but it does not provide any explanation of how complying with “one person, one *171vote” deters or identifies a gerrymander that unconstitutionally discriminates against a cognizable group of voters. While that standard affords some protection to the voting rights of individuals, “it protects groups only indirectly at best,” Karcher v. Daggett, 462 U. S., at 752 (Stevens, J., concurring), even when the group’s identity is determined solely by reference to the fact that its members reside in a particular voting district. “One person, one vote” alone does not protect the voting rights of a group made up of persons affiliated with a particular political party who seek to achieve representation through their combined voting strength. Thus, the facts that the legislature permitted each Democratic voter to cast his or her one vote, erected no direct barriers to Democratic voters’ exercise of the franchise, and drew districts of equal population, are irrelevant to a claim that district lines were drawn for the purpose and with the effect of substantially debasing the strength of votes cast by Democrats as a group.9
The final and most basic flaw in the plurality’s opinion is its failure to enunciate any standard that affords guidance to legislatures and courts.10 Legislators and judges are left to *172wonder whether compliance with “one person, one vote” completely insulates a partisan gerrymander from constitutional scrutiny, or whether a fairer but as yet undefined standard applies. The failure to articulate clear doctrine in this area *173places the plurality in the curious position of inviting further litigation even as it appears to signal the “constitutional green light”11 to would-be gerrymanderers.
h-H t — t h — H
In Karcher v. Daggett, Justice Stevens, echoing the decision in Reynolds v. Sims, described factors that I believe properly should guide both legislators who redistrict and judges who test redistricting plans against constitutional challenges. 462 U. S., at 753-761. The most important of these factors are the shapes of voting districts and adherence to established political subdivision boundaries.12 Other relevant considerations include the nature of the legislative procedures by which the apportionment law was adopted and legislative history reflecting contemporaneous legislative goals. To make out a case of unconstitutional partisan gerrymandering, the plaintiff should be required to offer proof concerning these factors, which bear directly on the fairness of a redistricting plan, as well as evidence concerning population disparities and statistics tending to show vote dilution. No one factor should be dispositive.13
*174In this case, appellees offered convincing proof of the ease with which mapmakers, consistent with the “one person, one vote” standard, may design a districting plan that purposefully discriminates against political opponents as well as racial minorities. Computer technology now enables gerrymandered to achieve their purpose while adhering perfectly to the requirement that districts be of equal population. Relying on the factors correctly described by Justice Stevens in Karcher v. Daggett, the District Court carefully reviewed appellees’ evidence and found that the redistricting law was intended to and did unconstitutionally discriminate against Democrats as a group. We have held that a district court’s ultimate determination that a redistricting plan was “being maintained for discriminatory purposes,” as well as its “subsidiary findings of fact,” may not be set aside by a reviewing court unless they are clearly erroneous. Rogers v. Lodge, 458 U. S. 613, 622-623 (1982); see, e. g., White v. Regester, 412 U. S. 755, 769-770 (1973). The plurality ignores these precedents.14 The plurality also disregards the various factors discussed by the District Court as adequate indicia of unconstitutional gerrymandering.
*175A
A court should look first to the legislative process by which the challenged plan was adopted. Here, the District Court found that the procedures used in redistricting Indiana were carefully designed to exclude Democrats from participating in the legislative process. In February 1981, both Houses of the General Assembly passed reapportionment bills with no substantive content and referred them to the other chamber where conflicting amendments were made. The purpose of this process was to send “vehicle bills” to a Conference Committee whose task was to apportion representation. Four conferees and four advisers served on the Committee. The conferees, all Republicans, were responsible for designing the voting districts and were entitled to vote on the result of their own efforts. The advisers, Democrats, were excluded from the mapmaking process and were given no Committee vote. 603 F. Supp., at 1483.
The legislative process consisted of nothing more than the majority party’s private application of computer technology to mapmaking. The Republican State Committee engaged the services of a computer firm to aid the conferees in their task. Id., at 1483-1484. According to the Conference Committee Chairman, the only data used in the computer program were precinct population, race of precinct citizens, precinct political complexion, and statewide party voting trends. Access to the mapmaking process was strictly limited. No member of the Democratic Party and no member of the public was provided with any of the information used in or generated by the computer program. When questioned about the lack of minority party participation in the redistricting process, the Chairman of the Conference Committee stated that the Democrats would “have the privilege to offer a minority map. But I will advise you in advance that it will not be accepted.” Id., at 1484.
*176Republicans promised to hold public hearings on redistricting. No hearing was held during the mapmaking process, the only time during which voters’ views could be expected to influence their legislators. Ibid. Two days before the end of the General Assembly’s regular session, during the first and only public hearing on reapportionment, the Conference Committee revealed for the first time the result of its map-making effort. This timing gave the Democrats but 40 hours in which to review the districting of more than 4,000 precincts. Ibid. On the last day of the session, April 30, 1981, the Conference Committee report was introduced for a vote and was adopted by party line vote in both Houses of the General Assembly.15 Ibid.
B
Next, the District Court found that the maps “conspicuously ignore[d] traditional political subdivisions, with no concern for any adherence to principles of community interest.” Id., at 1493. The court carefully described how the mapmakers carved up counties, cities, and even townships in their effort to draw lines beneficial to the majority party. Many districts meander through several counties, picking up a number of townships from each.16 The District Court explained why this failure to honor county boundaries could be expected to have a detrimental impact on citizens’ exercise of their vote. In Indiana, the county government is the seat of local affairs. Id., at 1494. The redistricting dissects counties into strange shapes lacking in common interests, on one occasion even placing the seat of one county in a voting dis*177trict composed of townships from other counties. Id., at 1487; see House Districts 45, 46, infra. Under these conditions, the District Court expressly found that “the potential for voter disillusion and nonparticipation is great,” as voters are forced to focus their political activities in artificial electoral units. 603 F. Supp., at 1494. Intelligent voters, regardless of party affiliation, resent this sort of political manipulation of the electorate for no public purpose.
Deposition testimony of the Chairman of the Conference Committee revealed that the mapmakers gave no consideration to the interests of communities. In the Chairman’s view, the concept of honoring community interests meant only that mapmakers should refuse to divide a small, suburban community. The shapes of the voting districts and the manner in which the districts divide established communities, from the county to the township level, illustrate that community interests were ignored by appellants. As the District Court observed, for example, “it is difficult to conceive the interests shared by blacks in Washington Township and white suburbanites in Hamilton and Boone Counties, or the shared interest of Allen and Noble County farmers with residents of downtown Fort Wayne.” Id., at 1487.
C
In addition to the foregoing findings that apply to both the House and Senate plans, the District Court also noted the substantial evidence that appellants were motivated solely by partisan considerations. Id., at 1484. There is no evidence that the public interest in a fair electoral process was given any consideration by appellants. Indeed, as noted above, the mapmakers’ partisan goals were made explicitly clear by contemporaneous statements of Republican leaders who openly acknowledged that their goal was to disadvantage Democratic voters. As one Republican House member concisely put it, “[t]he name of the game is to keep *178us in power.”17 NAACP Plaintiffs’ Exhibit 232 (Indianapolis Star, Mar. 22, 1981, section 2, p. 3). When the plan was completed, Republican leaders announced that the House map was designed to yield 56 “safe” Republican seats and 30 Democratic seats, with the remainder being “tossups.” NAACP Plaintiffs’ Exhibit 242 (Post-Tribune, Gary, Ind., Apr. 29,1981, p. 1). Republicans expected that their Senate map would regularly produce 30 Republican seats and 8 to 10 Democratic seats so that Republicans would maintain their grip on the Senate even if Democrats won the remaining seats. NAACP Plaintiffs’ Exhibit 241 (Post-Tribune, Gary, Ind., Apr. 29, 1981, p. 1). In short, the record unequivocally demonstrates that in 1981 the Republican-dominated General Assembly deliberately sought to design a redistricting plan under which members of the Democratic Party would be deprived of a fair opportunity to win control of the General Assembly at least until 1991, the date of the next redistricting.
IV
A
I turn now to the District Court’s findings with respect particularly to the gerrymandering of the House districts. The court found that the plan contained voting districts whose irrational shapes called for justification. E. g., House Districts 20, 22, 25, 45, 46, 48, 62, 66, 70, 73. The findings concerning the district configurations reflect the panel’s familiarity with Indiana geography and the particular characteristics of the State’s political subdivisions. As the District Court noted, the voter confusion generated by irrational dis*179trict boundaries is exacerbated in this case by the fact that the lines in the House plan were drawn independently of those in the Senate plan. 603 F. Supp., at 1484-1485. When the Senate voting districts are overlaid on the House districts, the potential for voter confusion becomes readily apparent as lines and districts intersect in a crazy quilt.18
The District Court carefully considered the multimember districts contained in the House plan and found that they were intentionally employed to minimize Democratic voting power. This Court has expressly recognized that “[a] dis-tricting plan may create multimember districts perfectly acceptable under equal population standards, but invidiously discriminatory because they are employed ‘to minimize or cancel out the voting strength of racial or political elements of the voting population.’” Gaffney v. Cummings, 412 U. S., at 751 (quoting Fortson v. Dorsey, 379 U. S. 433, 439 (1965)).19 In this case, invidious purpose may be inferred from the mapmakers’ selection of areas to be divided into multimember districts. These districts appear in some areas where they had been used previously and not in others, in some urban areas and not in others, and in some areas where *180their use required combining rural townships with urban areas from another county.20 The only discernible pattern is the appearance of these districts in areas where their winner-take-all aspects can best be employed to debase Democratic voting strength. The District Court determined that the multimember districts diluted Democratic voting strength by “stacking” Democrats into districts “where their majority would be overwhelming” and by fragmenting populations of Democratic voters among other districts where their voting strength would be reduced. 603 F. Supp., at 1488-1489, 1494. For example, the mapmakers split Fort Wayne, a city with a demonstrated tendency to vote for Democratic candidates, and associated each of the halves with areas from outlying counties whose residents had a pattern of voting for Republican candidates. Id., at 1488, 1494; see House Districts 19, 20. Similarly, the redistricting of Marion County presents a clear example of dilution of Democrats’ voting strength through the use of multimember districts. Though population figures entitled the county to elect exactly 14 House members, the mapmakers decided to tack on portions of two neighboring counties in order artificially to create a population base entitled to elect 15 representatives. Then, they carved that artificial geographical unit into five three-member districts whose irregular shapes were designed to fence Democrats into one heavily Democratic district and scatter pockets of Democratic strength among the other four districts. Id., at 1487, 1489; see House Districts 48, 49, 50, 51, 52.21
*181Appellees further demonstrated through a statistical showing that the House plan debased the effectiveness of their votes. In 1982, all 100 House seats were up for election. *182Democratic candidates received about 51.9 percent of the vote, and Republican candidates received about 48.1 percent. Forty-three Democratic representatives were elected; 57 Republicans were elected.22 Appellees offered startling statistics with respect to House results in Marion and Allen Counties, two areas in which multimember districts were used. In these counties, Democratic candidates earned 46.3 percent of the vote, but won only 3 of 21 House seats. As the District Court observed, “such a disparity speaks for itself.” Id., at 1489.23
B
Since half of the Senate membership is up for election every two years, the only election results under the challenged plan available at trial related to 25 of the 50 Senate seats. Those results showed that, of the seats up for election in 1982, Democrats were elected to 13 seats and Republicans to 12. Democratic candidates earned about 53.1 percent of the vote, and Republicans received about 46.9 percent. At trial, it was appellees’ contention that most of the Senate seats won by Democrats in 1982 were “safe” Democratic seats so that their party’s success at the polls in that year was fully consistent with the statewide Republican gerrymander. This contention is borne out by the results of the 1984 Senate election. In that election, Democratic candidates received 42.3 percent of the vote, and Republicans 57.7 *183percent. Yet, of the 25 Senate positions up for election, only 7 were captured by Democrats.24
*184C
The District Court found, and I agree, that appellants failed to justify the discriminatory impact of the plan by showing that the plan had a rational basis in permissible neutral criteria. Appellants’ primary justification was that the plan comports with the principle of “one person, one vote.” Their plan did adhere to that objective, with population deviations between House districts of 1.05 percent and between Senate districts of 1.15 percent. But reliance on “one person, one vote” does not sufficiently explain or justify the discrimination the plan inflicted on Democratic voters as a group. The District Court expressly found that the irregular district shapes could not be justified on the basis of population distribution. Id., at 1494. Nor does adherence to “one person, one vote” excuse the mapmakers’ failure to honor established political or community boundaries. It does not excuse the irrational use of multimember districts, with their devastating impact on the voting strength of Democrats. The only other justification offered by appellants, for which the District Court found some support as a contemporaneous goal, was that the mapmakers sought to maintain “the black representation in the General Assembly that existed prior to the new districting plan. ” But the court further determined that the impact of the redistricting fell most harshly on black voters who predominantly are Democrats. Id., at 1488, 1489-1490. None of these critical findings was found by the plurality today to be clearly erroneous.
V
In conclusion, I want to make clear the limits of the standard that I believe the Equal Protection Clause imposes on legislators engaged in redistricting. Traditionally, the determination of electoral districts within a State has been a matter left to the legislative branch of the state government. Apart from the doctrine of separation of powers and the federal system prescribed by the Constitution, federal *185judges are ill equipped generally to review legislative decisions respecting redistricting. As the Court’s opinion makes clear, however, our precedents hold that a colorable claim of discriminatory gerrymandering presents a justiciable controversy under the Equal Protection Clause. Federal courts in exercising their duty to adjudicate such claims should impose a heavy burden of proof on those who allege that a redistricting plan violates the Constitution. In light of Baker v. Carr, Reynolds v. Sims, and their progeny, including such comparatively recent decisions as Gaffney v. Cummings, this case presents a paradigm example of unconstitutional discrimination against the members of a political party that happened to be out of power. The well-grounded findings of the District Court to this effect have not been, and I believe cannot be, held clearly erroneous.
Accordingly, I would affirm the judgment of the District Court.25
[Maps of the Indiana House and Senate Districts follow this page.]
*187[[Image here]]
*188[[Image here]]
This opinion uses the term “redistricting” to refer to the process by which state legislators draw the boundaries of voting districts. The terms “redistricting,” “apportionment,” and “reapportionment” frequently are used interchangeably. Backstrom, Robins, & Eller, Issues in Gerrymandering: An Exploratory Measure of Partisan Gerrymandering Applied to Minnesota, 62 Minn. L. Rev. 1121, 1121, n. 1 (1978); Grofman, Criteria for Districting: A Social Science Perspective, 33 UCLA L. Rev. 77, 78, n. 6 (1985). Technically, the words “apportionment” and “reapportionment” apply to the “allocation of a finite number of representatives among a fixed number of pre-established areas,” while “districting” and “redistricting” refer to the drawing of district lines. Backstrom, Robins, & Eller, supra, at 1121, n. 1; see Grofman, supra, at 78, n. 6.
In the District Court, appellees’ lawsuit was consolidated with a suit brought by the Indiana NAACP. The plaintiffs in the NAACP suit argued that the redistrieting intentionally fragmented concentrations of black voters in violation of the Fourteenth and Fifteenth Amendments, and of § 2 of the Voting Rights Act of 1965, 42 U. S. C. § 1973. The District Court determined that the plan discriminated against black voters, not because of their race, but because blacks had a demonstrated and overwhelming tendency to vote as a bloc for Democratic candidates. Indeed, the District Court explicitly found that the “disadvantaging effect of the plan’s multi-member districts falls particularly hard and harsh upon black voters in the state.” 603 F. Supp. 1479, 1488 (SD Ind. 1984). Rather than taking a cross-appeal challenging the District Court’s rejection of their constitutional and statutory claims, the NAACP plaintiffs have filed a brief in this Court urging affirmance of the District Court’s judgment that the plan unconstitutionally discriminates against Democratic voters as a group and against blacks as members of that group.
Webster’s Third New International Dictionary (unabridged ed. 1961) defines “gerrymander” as “to divide (a territorial unit) into election districts in an unnatural and unfair way with the purpose of giving one political party an electoral majority in a large number of districts while concentrating the voting strength of the opposition in as few districts as possible.” The term “gerrymander” was coined by combining the last name of Elbridge Gerry with the word “salamander” in order to describe the “fancied resemblance to a salamander ... of the irregularly shaped outline of an election district in northeastern Massachusetts that had been formed for partisan purposes in 1812 during Gerry’s governorship” of that State. Ibid. Though many of the voting districts appearing in the plans challenged here have bizarre shapes, House District 66 perhaps most closely resembles a salamander. See the redistricting maps appended to this opinion. These maps were reproduced from trial exhibits provided by the parties.
As the Court properly explains, our prior decisions make clear that an equal protection challenge to redistricting does not present a nonjusticiable political question. See Baker v. Carr, 369 U. S. 186 (1962); Reynolds v. Sims, 377 U. S. 533 (1964); Gaffney v. Cummings, 412 U. S. 735 (1973). Accordingly, I join Part II of Justice White’s opinion.
The doctrine of “one person, one vote” originally was regarded as a means to prevent discriminatory gerrymandering since “opportunities for gerrymandering are greatest when there is freedom to construct unequally populated districts.” Kirkpatrick v. Preisler, 394 U. S. 526, 534, n. 4 (1969). Advances in computer technology achieved since the doctrine was announced have drastically reduced its deterrent value by permitting political cartographers to draw districts of equal population that intentionally discriminate against cognizable groups of voters. See Karcher v. Daggett, 462 U. S. 725, 752, n. 10 (1983) (Stevens, J., concqrring). For “one person, one vote” to serve its intended purpose of implementing the constitutional mandate of fair and effective representation, therefore, consideration also must be given to other neutral factors.
In decisions concerning congressional redistricting, the Court has focused its attention almost exclusively on whether a challenged plan satisfies “one person, one vote.” See Karcher v. Daggett, supra; White v. Weiser, 412 U. S. 783 (1973); Kirkpatrick v. Preisler, supra. In cases involving state legislative redistricting, such as the case before us today, the Court has refused to limit a legislature to the single goal of precise population equality. Gaffney v. Cummings, 412 U. S., at 745; Mahan v. Howell, 410 U. S. 315, 322-325 (1973).
The plurality correctly concludes that a redistricting plan is not unconstitutional merely because the plan makes it more difficult for a group of voters to elect the candidate of its choice or merely because the plan does not provide proportional representation. Ante, at 132. While the “Equal Protection Clause confers a substantive right to participate in elections on an equal basis with other qualified voters,” Mobile v. Bolden, 446 U. S. 55, 77 (1980) (plurality opinion), the Constitution does not guarantee proportional representation or protect any group from defeat at the polls, id., at 77-80; see White v. Regester, 412 U. S. 755, 765-766 (1973). But the plurality leaps from that conclusion to the assumption that “[a]n individual or a group of individuals who votes for a losing candidate is usually deemed to be adequately represented by the winning candidate and to have as much opportunity to influence that candidate as other voters in the district.” Ante, at 132. Thus, the plurality apparently believes that effects on election results are of little import, as long as the losers have some access to their representatives. Though effects on election results do not suffice to establish an unconstitutional gerrymander, they certainly are *170relevant to such a claim, and they may suffice to show that the claimants have been injured by the redistricting they challenge.
The District Court found: “Control of the General Assembly is crucial to a political party for a number of reasons. The majority party elects the Speaker of the House, a person who wields considerable power in the assigning of bills to committees, the conduct of the actual legislative sessions, and is empowered, under legislative rules, to prevent bills from reaching the floor for debate or vote. Similarly, the majority party elects floor leaders in both houses who control the flow of legislation, the assignment of members to committees, and the appointment of committee chairmen. All of these powers are important to the achievement of a party’s legislative goals. There is little doubt that the minority party plays a less substantial role in the drafting and enactment of legislation.” 603 F. Supp., at 1483.
As was said in the context of a constitutional challenge by black voters to an at-large voting scheme, “[t]he absence of official obstacles to registration, voting, and running for office heretofore has never been deemed to insulate an electoral system” from constitutional attack. Mobile v. Bolden, supra, at 102 (White, J., dissenting).
The plurality describes its standard as requiring a “threshold” showing that the “electoral system is arranged in a manner that will consistently degrade a voter’s or a group of voters’ influence in the political process as a whole.” Ante, at 132; see ante, at 134, n. 14. Plaintiffs apparently can meet the plurality’s “threshold” only after a number of elections have been held under the challenged plan. Ante, at 135-136. At one point, the plurality acknowledges that this formulation is “somewhat different” from any standard we have previously used to test an electoral plan against an equal protection challenge, ante, at 132, and also takes pains to say that its opinion here does not suggest any “alteration of the standards developed” for evaluating racial challenges, ante, at 132, n. 13; see ante, at 134, n. 14. Curiously, the plurality then goes on to claim that its standard is consistent *172with that used when a racial group challenges an electoral scheme, ante, at 139-140, and with our “equal protection cases generally,” ante, at 141. This claim is simply incorrect.
Our cases have construed the Equal Protection Clause to require proof of intentional discrimination, placing the burden on plaintiffs to trace the “ ‘invidious quality of a law claimed to be racially discriminatory ... to a racially discriminatory purpose.’” Rogers v. Lodge, 458 U. S. 613, 616 (1982), quoting Washington v. Davis, 426 U. S. 229, 240 (1976). In none of those cases was the Court willing to assume discriminatory intent, as the plurality suggests today is the proper course. Ante, at 141. While the plurality correctly observes that our prior decisions have held that disproportionate election results alone do not violate the Constitution, it erroneously suggests that those holdings flowed solely from the “perception that the power to influence the political process is not limited to winning elections.” Ante, at 131-132. The plurality wholly ignores the basic problem underlying all of those prior decisions, namely, that the plaintiffs came into court with no direct proof of discriminatory intent. In those cases, the Court concluded that proof of discriminatory effect, including disproportionate election results, if serious enough, could give rise to an inference of purposeful discrimination. See Rogers v. Lodge, supra, at 618. As Justice White has explained, the Court’s decisions in both White v. Regester, 412 U. S. 755 (1973), and Whitcomb v. Chavis, 403 U. S. 124 (1971), rested on the proposition that the requisite “invidious discriminatory purpose” can be inferred from proof of “objective factors” concerning discriminatory effect. Mobile v. Bolden, 446 U. S., at 95; see id., at 94-97, 102-103 (dissenting opinion); see also White v. Regester, supra, at 765 (multimember districts are unconstitutional where “used invidiously to cancel out” racial groups’ voting strength). I cannot agree, as the plurality suggests, that a standard requiring proof of “heightened effect,” where invidious intent has been established directly, has support in any of our cases, or that an equal protection violation can be established “only where a history (actual or projected) of disproportionate results appears.” Ante, at 139. If a racial minority established that the legislature adopted a redistricting law for no purpose other than to disadvantage that group, the plurality’s new and erroneous standard would require plaintiffs to wait for the results of several elections, creating a history of discriminatory effect, before they can challenge the law in court. Ante, at 135-136.
Edwards, The Gerrymander and “One Man, One Vote,” 46 N. Y. U. L. Rev. 879, 880 (1971).
In some cases, proof of grotesque district shapes may, without more, provide convincing proof of unconstitutional gerrymandering. In addition to the maps appended to this opinion, see the redistricting maps appended to the Court’s opinions in Gomillion v. Lightfoot, 364 U. S. 339, 348 (1960), and in Karcher v. Daggett, 462 U. S., at 744.
Groups may consistently fail to elect representatives under a perfectly neutral election scheme. Thus, a test that turns only on election results, as the plurality’s standard apparently does, likely would identify an unconstitutional gerrymander where none existed. The test that I would adopt requires consideration of all the circumstances surrounding the plan, including factors initially identified in Reynolds v. Sims, to determine if a constitutional violation has occurred. Since democracy can work well and fairly only when citizens have an opportunity to become familiar with their voting districts, where they must focus their political activities, district boundaries must have some rationality. Confusion inevitably follows, as clearly will be the ease in Indiana, when a citizen finds himself or herself *174forced to associate with several artificial communities, depending on which office is on the ballot. Thus, irrational lines themselves affect the ability of all voters to exercise their political influence, with disproportionate election results illustrating the effect of a plan that was deliberately designed to serve no purpose other than to minimize the voting strength of a disfavored group.
The plurality ignores the “clearly erroneous” standard of Rule 52(a), by saying that it has not rejected any of the District Court’s findings of fact, but has “merely . . . disregarded those that were irrelevant in this case and held insufficient those that inadequately supported the District Court’s ultimate legal conclusions.” Ante, at 142, n. 20. In a gerrymandering ease the facts as to how, where, and why the legislature drew the district boundaries are at the heart of the equal protection violation. Beyond stating that appellees’ statistical proof failed to satisfy its proposed threshold, the plurality makes no effort to explain its flat assertions that the District Court’s careful findings were “irrelevant” or “insufficient.”
Presumably as a result of the haste with which the redistricting law was pushed through the General Assembly, parts of the State were “wholly omitted in the 1981 legislation.” 603 F. Supp., at 1484. In the 1982 legislative session, therefore, amendments were passed to assign the omitted areas to voting districts. Ibid.
E. g., House Districts 20, 22, 25, 28, 42, 45, 46, 55, 57, 62, 66, 70, 73, 74; Senate Districts 7, 24, 37, 39, 45, 47. See the redistricting maps appended to this opinion.
Evidence of partisan sparring during the redistrieting process, of course, is not sufficient to establish an equal protection violation or to show that the legislature pursued no legitimate objectives in adopting the plan. But such evidence is probative of contemporaneous legislative goals, adding support to the objective facts showing that the legislature adopted the plan for the sole purpose of disadvantaging members of the political party that happened to be out of power.
Since the Indiana House of Representatives has 100 members, and the Senate has 50, the mapmakers readily could have designed a “nested” plan, that is, a plan that included “two House districts within one Senate district.” 603 F. Supp., at 1484-1485. By permitting voters readily to identify their voting districts and corresponding representatives, a nested plan can be expected to foster voter participation. See Grofman, 33 UCLA L. Rev., at 88, 92. Instead, as the District Court observed, the mapmakers drew House districts that were “not at all relevant to the Senate districts.” 603 F. Supp., at 1484.
In the context of racial gerrymandering claims, the Court has refused to adopt a per se rule barring the use of multimember districts. White v. Regester, 412 U. S., at 765. But the Court has repeatedly recognized that the characteristics of multimember districts, namely, their tendency to submerge the voting strength of the minority by allowing the majority to capture all of the district’s assigned seats, make them a ready means for legislative discrimination against racial groups or political opponents. E. g., Rogers v. Lodge, 458 U. S., at 616.
The multimember districts are House Districts 7, 9, 10, 11, 12, 14, 15, 19, 20, 31, 48, 49, 50, 51, 52, 75.
The District Court found that the multimember districts employed in Marion County were “particularly suspect with respect to compactness.” 603 F. Supp., at 1487. Of all the districts in the challenged plan, the court determined that House District 48 “presents the most grievous example of the political cartographer’s handiwork in this case.” Ibid. That district “forms the letter ‘C’ around the central city of Indianapolis” and “includes portions of the urban southwestside of the city, the airport and suburban *181area around Ben Davis High School on the west side, and the Meridian Hills area at the northern part of the county.” Ibid. The court expressly determined that, even though House District 48 satisfies “one person, one vote,” there was “simply no conceivable justification for this kind of district.” Ibid.
The following map, taken from an exhibit provided by the parties, shows this grotesque gerrymandering. The legislature first proceeded to disregard Marion County’s boundary lines, which essentially form a square, and then carved the area it created into oddly shaped multimember districts.
[[Image here]]
The District Court also noted the discriminatory purpose served by the Marion County House Districts, including District 48: “[T]he powerful Marion County delegation forced neighboring counties to cede turf to permit a preservation of the multi-member districts which had consistently returned Republicans to the Statehouse.” Id., at 1487, n. 1. Moreover, as appellees’ statistical showing of vote dilution plainly demonstrates, these gerrymandered districts had a discriminatory impact on the votes of Democrats as a group.
In the 1984 election, Democratic candidates earned approximately 44 percent of the vote, and Republicans earned approximately 56 percent. Thirty-nine Democrats were elected to the House, and 61 Republicans were elected. The figures for the 1984 elections cited in this opinion were provided by the Elections Research Center, Washington, D. C. A supplemental statement filed by appellants in the District Court following trial also quoted some of the 1984 election results, including the fact that in 1984 the Democratic candidate for Governor won 48 percent of the vote.
The 1984 House election in the Marion and Allen County House districts reflected a similar disparity, when Republicans again captured 18 of the 21 House seats and the Democrats 3 despite the fact that Democratic candidates earned approximately 38 percent of the vote in these counties.
The District Court’s discussion of district shapes focused primarily on the House plan. As the following map of the Senate districts in the Marion County area illustrates, the Senate plan also contains districts with unusual shapes. Although the population of Marion County, whose boundary lines form a square, was entitled to elect exactly seven Senators, 603 F. Supp., at 1487, n. 1, the mapmakers ignored both that population figure and the county boundaries, and created eight wholly irrational voting districts. As one Democratic voter remarked when the Senate plan was unveiled: “People who live near the [district line separating Senate districts 33 and 34] are going to need an Indian guide and a compass to figure out which district they’re in.” NAACP Plaintiffs’ Exhibit 252 (Indianapolis Star, May 10, 1981, section 2, p. 3).
[[Image here]]
As is evident from the several opinions filed today, there is no “Court” for a standard that properly should be applied in determining whether a challenged redistricting plan is an unconstitutional partisan political gerrymander. The standard proposed by the plurality is explicitly rejected by two Justices, and three Justices also have expressed the view that the plurality’s standard will “prove unmanageable and arbitrary.” Ante, at 155 (O’Connor, J., joined by Burger, C. J., and Rehnquist, J., concurring in judgment).