concurring in part and dissenting in part.
I join Parts II-A and III of the Court’s opinion. I also join Parts I and II of Justice Stevens’ opinion concurring in part and dissenting in part.
For one thing, the timing of the redistricting (between census periods), the radical departure from traditional boundary-drawing criteria, and the other evidence to which Justice Stevens refers in Parts I and II of his opinion make clear that a “desire to maximize partisan advantage” was the “sole purpose behind the decision to promulgate Plan 1374C.” Ante, at 458. Compare, e. g., App. 176-178; ante, at 452-455, 458-459 (opinion of Stevens, J.), with Vieth v. Jubelirer, 541 U. S. 267, 366-367 (2004) (Breyer, J., dissenting).
For another thing, the evidence to which Justice Stevens refers in Part III of his opinion demonstrates that the *492plan’s effort “to maximize partisan advantage,” ante, at 458, encompasses an effort not only to exaggerate the favored party’s electoral majority but also to produce a majority of congressional representatives even if the favored party receives only a minority of popular votes. Compare ante, at 465-468 (opinion of Stevens, J.), App. 55 (plaintiffs’ expert), and id., at 216 (State’s expert), with Vieth, supra, at 360 (Breyer, J., dissenting).
Finally, because the plan entrenches the Republican Party, the State cannot successfully defend it as an effort simply to neutralize the Democratic Party’s previous political gerrymander. Nor has the State tried to justify the plan on nonpartisan grounds, either as an effort to achieve legislative stability by avoiding legislative exaggeration of small shifts in party preferences, see Vieth, 541 U. S., at 359 (same), or in any other way.
In sum, “the risk of entrenchment is demonstrated,” “partisan considerations [have] rendered] the traditional district-drawing compromises irrelevant,” and “no justification other than party advantage can be found.” Id., at 367 (same). The record reveals a plan that overwhelmingly relies upon the unjustified use of purely partisan line-drawing considerations and which will likely have seriously harmful electoral consequences. Ibid. For these reasons, I believe the plan in its entirety violates the Equal Protection Clause.
Chief Justice Roberts, with whom Justice Alito joins, concurring in part, concurring in the judgment in part, and dissenting in part.
I join Parts I and IV of the plurality opinion. With regard to Part II, I agree with the determination that appellants have not provided “a reliable standard for identifying unconstitutional political gerrymanders.” Ante, at 423. The question whether any such standard exists—that is, whether a challenge to a political gerrymander presents a justiciable case or controversy—has not been argued in these cases. I therefore take no position on that question, which *493has divided the Court, see Vieth v. Jubelirer, 541 U. S. 267 (2004), and I join the Court’s disposition in Part II without specifying whether appellants have failed to state a claim on which relief can be granted, or have failed to present a justiciable controversy.
I must, however, dissent from Part III of the Court’s opinion. According to the District Court’s factual findings, the State’s drawing of district lines in south and west Texas caused the area to move from five out of seven effective Latino opportunity congressional districts, with an additional district “moving” in that direction, to six out of seven effective Latino opportunity districts. See Session v. Perry, 298 F. Supp. 2d 451, 489, 503-504 (ED Tex. 2004) (per curiam). The end result is that while Latinos make up 58% of the citizen voting-age population in the area, they control 85% (six of seven) of the districts under the State’s plan.
In the face of these findings, the majority nonetheless concludes that the State’s plan somehow dilutes the voting strength of Latinos in violation of § 2 of the Voting Rights Act of 1965. The majority reaches its surprising result because it finds that Latino voters in one of the State’s Latino opportunity districts—District 25—are insufficiently compact, in that they consist of two different groups, one from around the Rio Grande and another from around Austin. According to the majority, this may make it more difficult for certain Latino-preferred candidates to be elected from that district—even though Latino voters make up 55% of the citizen voting-age population in the district and vote as a bloc. Id., at 492, n. 126, 503. The majority prefers old District 23, despite the District Court determination that new District 25 is “a more effective Latino opportunity district than Congressional District 23 had been.” Id., at 503; see id., at 489,498-499. The District Court based that determination on a careful examination of regression analysis showing that “the Hispanic-preferred candidate [would win] every primary and general election examined in District 25,” id., at 503 (emphasis added), compared to the only partial success *494such candidates enjoyed in former District 23, id., at 488, 489, 496.
The majority dismisses the District Court’s careful fact-finding on the ground that the experienced judges did not properly consider whether District 25 was “compact” for purposes of § 2. Ante, at 430-431. But the District Court opinion itself clearly demonstrates that the court carefully considered the compactness of the minority group in District 25, just as the majority says it should have. The District Court recognized the very features of District 25 highlighted by the majority and unambiguously concluded, under the totality of the circumstances, that the district was an effective Latino opportunity district, and that no violation of § 2 in the area had been shown.
Unable to escape the District Court’s factfinding, the majority is left in the awkward position of maintaining that its theory about compactness is more important under § 2 than the actual prospects of electoral success for Latino-preferred candidates under a State’s apportionment plan. And that theory is a novel one to boot. Never before has this or any other court struck down a State’s redistricting plan under § 2, on the ground that the plan achieves the maximum number of possible majority-minority districts, but loses on style points, in that the minority voters in one of those districts are not as “compact” as the minority voters would be in another district were the lines drawn differently. Such a basis for liability pushes voting rights litigation into a whole new area—an area far removed from the concern of the Voting Rights Act to ensure minority voters an equal opportunity “to elect representatives of their choice.” 42 U. S. C. § 1973(b).
I
Under § 2, a plaintiff alleging “a denial or abridgement of the right of [a] citizen of the United States to vote on account of race or color,” § 1973(a), must show, “based on the totality of circumstances,”
*495“that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a). .. in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.” § 1973(b).
In Thornburg v. Gingles, 478 U. S. 30 (1986), we found that a plaintiff challenging the State’s use of multimember districts could meet this standard by showing that replacement of the multimember district with several single-member districts would likely provide minority voters in at least some of those single-member districts “the ability ... to elect representatives of their choice.” Id., at 48. The basis for this requirement was simple: If no districts were possible in which minority voters had prospects of electoral success, then the use of multimember districts could hardly be said to thwart minority voting power under § 2. See ibid. (“Minority voters who contend that the multimember form of districting violates § 2 must prove that the use of a multimember electoral structure operates to minimize or cancel out their ability to elect their preferred candidates”).
The next generation of voting rights litigation confirmed that “manipulation of [single-member] district lines” could also dilute minority voting power if it packed minority voters in a few districts when they might control more, or dispersed them among districts when they might control some. Voinovich v. Quilter, 507 U. S. 146, 153-154 (1993). Again the basis for this application of Gingles was clear: A configuration of district lines could only dilute minority voting strength if under another configuration minority voters had better electoral prospects. Thus in cases involving single-member districts, the question was whether an additional majority-minority district should be created, see Abrams v. Johnson, 521 U. S. 74, 91-92 (1997); Growe v. Emison, 507 U. S. 25, 38 (1993), or whether additional influence districts *496should be created to supplement existing majority-minority districts, see Voinovich, supra, at 154.
We have thus emphasized, since Gingles itself, that a §2 plaintiff must at least show an apportionment that is likely to perform better for minority voters, compared to the existing one. See 478 U. S., at 99 (O’Connor, J., concurring in judgment) (“[T]he relative lack of minority electoral success under a challenged plan, when compared with the success that would be predicted under the measure of undiluted minority voting strength the court is employing, can constitute powerful evidence of vote dilution”). And unsurprisingly, in the context of single-member districting schemes, we have invariably understood this to require the possibility of additional single-member districts that minority voters might control.
Johnson v. De Grandy, 512 U. S. 997 (1994), reaffirmed this understanding. The plaintiffs in De Grandy claimed that, by reducing the size of the Hispanic majority in some districts, additional Hispanic-majority districts could be created. Id., at 1008. The State defended a plan that did not do so on the ground that the proposed additional districts, while containing nominal Hispanic majorities, would “lack enough Hispanic voters to elect candidates of their choice without cross-over votes from other ethnic groups,” and thus could not bolster Hispanic voting strength under § 2. Ibid.
In keeping with the requirement that a § 2 plaintiff must show that an alternative apportionment would present better prospects for minority-preferred candidates, the Court set out the condition that a challenge to an existing set of single-member districts must show the possibility of “creating more than the existing number of reasonably compact districts with a sufficiently large minority population to elect candidates of its choice.” Ibid. De Grandy confirmed that simply proposing a set of districts that divides up a minority population in a different manner than the State has chosen, *497without a gain in minority opportunity districts, does not show vote dilution, but “only that lines could have been drawn elsewhere.” Id., at 1015.
Here the District Court found that six Latino-majority districts were all that south and west Texas could support. Plan 1374C provides six such districts, just as its predecessor did. This fact, combined with our precedent making clear that § 2 plaintiffs must show an alternative with better prospects for minority success, should have resulted in affirmance of the District Court decision on vote dilution in south and west Texas. See Gingles, supra, at 79 (“[T]he clearly-erroneous test of [Federal Rule of Civil Procedure] 52(a) is the appropriate standard for appellate review of a finding of vote dilution. . . . [Wjhether the political process is equally open to minority voters ... is peculiarly dependent upon the facts” (internal quotation marks omitted)); Rogers v. Lodge, 458 U. S. 613, 622, 627 (1982).
The majority avoids this result by finding fault with the District Court’s analysis of one of the Latino-majority districts in the State’s plan. That district—District 25—is like other districts in the State’s plan, like districts in the predecessor plan, and like districts in the plaintiffs’ proposed seven-district plan, in that it joins population concentrations around the border area with others closer to the center of the State. The District Court explained that such “ 'bacon-strip’ ” districts are inevitable, given the geography and demography of that area of the State. Session, 298 F. Supp. 2d, at 486-487, 490, 491, n. 125, 502.
The majority, however, criticizes the District Court because its consideration of the compactness of District 25 under §2 was deficient. According to the majority,
“the court analyzed the issue only for equal protection purposes. In the equal protection context, compactness focuses on the contours of district lines to determine whether race was the predominant factor in drawing *498those lines. Under § 2, by contrast, the injury is vote dilution, so the compactness inquiry embraces different considerations.” Ante, at 433 (citation omitted).
This is simply an inaccurate description of the District Court’s opinion. The District Court expressly considered compactness in the §2 context. That is clear enough from the fact that the majority quotes the District Court’s opinion in elaborating on the standard of compactness it believes the District Court should have applied. See ante, at 424 (quoting Session, supra, at 502); ante, at 434 (quoting Session, supra, at 502). The very passage quoted by the majority about the different “ ‘needs and interests’ ” of the communities in District 25, ante, at 424, appeared in the District Court opinion precisely because the District Court recognized that those concerns “bear on the extent to which the new districts”—including District 25—“are functionally effective Latino opportunity districts, important to understanding whether dilution results from Plan 1374C,” Session, 298 F. Supp. 2d, at 502 (emphasis added); see also ibid. (noting different “needs and interests of Latino communities” in the “‘bacon-strip’” districts and concluding that “[t]he issue is whether these features mean that the newly-configured districts dilute the voting strength of Latinos” (emphasis added)).
Indeed, the District Court addressed compactness in two different sections of its opinion: in Part VI-C with respect to vote dilution under §2, and in Part VI-D with respect to whether race predominated in drawing district lines, for purposes of equal protection analysis. The District Court even explained, in considering in Part VI-C the differences between the Latino communities in the bacon-strip districts (including District 25) for purposes of vote dilution under § 2, how the same concerns bear on the plaintiffs’ equal protection claim, discussed in Part Vl-D. Id., at 502, n. 168. The majority faults the District Court for discussing “the relative smoothness of the district lines,” because that is only perti*499nent in the equal protection context, ante, at 432, but it was only in the equal protection context that the District Court mentioned the relative smoothness of district lines. See 298 F. Supp. 2d, at 506-508. In discussing compactness in Part VI-C, with respect to vote dilution under § 2, the District Court considered precisely what the majority says it should have: the diverse needs and interests of the different Latino communities in the district. Unlike the majority, however, the District Court properly recognized that the question under §2 was “whether these features mean that the newly-configured districts dilute the voting strength of Latinos.” Id., at 502.
The District Court’s answer to that question was unambiguous:
“Witnesses testified that Congressional Districts 15 and 25 would span colonias in Hidalgo County and suburban areas in Central Texas, but the witnesses testified, and the regression data show, that both districts are effective Latino opportunity districts, with the Hispaniepreferred candidate winning every primary and general election examined in District 25.” Id., at 503.
The District Court emphasized this point again later on:
“The newly-configured Districts 15, 25, 27, and 28 cover more territory and travel farther north than did the corresponding districts in Plan 1151C. The districts combine more voters from the central part of the State with voters from the border cities than was the case in Plan 1151C. The population data, regression analyses, and the testimony of both expert witnesses and witnesses knowledgeable about how politics actually works in the area lead to the finding that in Congressional Districts 25 and 28, Latino voters will likely control every primary and general election outcome.” Id., at 503-504.
I find it inexplicable how the majority can read these passages and state that the District Court reached its finding *500on the effectiveness of District 25 “without accounting for the detrimental consequences of its compactness problems.” Ante, at 442. The majority does “not question” the District Court’s parsing of the statistical evidence to reach the finding that District 25 was an effective Latino opportunity district. Ante, at 434. But the majority nonetheless rejects that finding, based on its own theory that “[t]he practical consequence of drawing a district to cover two distant, disparate communities is that one or both groups will be unable to achieve their political goals,” ibid., and because the finding rests on the “prohibited assumption” that voters of the same race will “think alike, share the same political interests, and will prefer the same candidates at the polls,” ante, at 433 (internal quotation marks omitted). It is important to be perfectly clear about the following, out of fairness to the District Court if for no other reason: No one has made any “assumptions” about how voters in District 25 will vote based on their ethnic background. Not the District Court; not this dissent. There was a trial. At trials, assumptions and assertions give way to facts. In voting rights cases, that is typically done through regression analyses of past voting records. Here, those analyses showed that the Latino candidate of choice prevailed in every primary and general election examined for District 25. See Session, 298 F. Supp. 2d, at 499-500. Indeed, a plaintiffs’ expert conceded that Latino voters in District 25 “have an effective opportunity to control outcomes in both primary and general elections.” Id., at 500. The District Court, far from “assum[ing]” that Latino voters in District 25 would “prefer the same candidate at the polls,” concluded that they were likely to do so based on statistical evidence of historic voting patterns.
Contrary to the erroneous statements in the majority opinion, the District Court judges did not simply “aggregate] ” minority voters to measure effectiveness. Ante, at 432. They did not simply rely on the “mathematical possibility” of minority voters voting for the same preferred *501candidate, ante, at 435, and it is a disservice to them to state otherwise. It is the majority that is indulging in unwarranted “assumption^]” about voting, contrary to the facts found at trial based on carefully considered evidence.
What is blushingly ironic is that the district preferred by the majority—former District 23—suffers from the same “flaw” the majority ascribes to District 25, except to a greater degree. While the majority decries District 25 because the Latino communities there are separated by “enormous geographical distance,” ibid., and are “hundreds of miles apart,” ante, at 441, Latino communities joined to form the voting majority in old District 23 are nearly twice as far apart. Old District 23 runs “from El Paso, over 500 miles, into San Antonio and down into Laredo. It covers a much longer distance than . . . the 300 miles from Travis to McAllen [in District 25].” App. 292 (testimony of T. Giberson); see id., at 314 (expert report of T. Giberson) (“[District 23 in any recent Congressional plan extends from the outskirts of El Paso down to Laredo, dipping into San Antonio and spanning 540 miles”). So much for the significance of “enormous geographical distance.” Or perhaps the majority is willing to “assume” that Latinos around San Antonio have common interests with those on the Rio Grande rather than those around Austin, even though San Antonio and Austin are a good bit closer to each other (less than 80 miles apart) than either is to the Rio Grande.*
*502The District Court considered expert evidence on projected election returns and concluded that District 25 would likely perform impeccably for Latino voters, better indeed than former District 23. See Session, 298 F. Supp. 2d, at 503-504, 488, 489, 496. The District Court also concluded that the other districts in Plan 1374C would give Latino voters a favorable opportunity to elect their preferred candidates. See id., at 499 (observing the parties’ agreement that Districts 16 and 20 in Plan 1374C “do clearly provide effective Latino citizen voting age population majorities”); id., at 504 (“Latino voters will likely control every primary and general election outcome” in District 28, and “every primary outcome and almost every general election outcome” in Districts 15 and 27, under Plan 1374C). In light of these findings, the District Court concluded that “compared to Plan 1151C ... Plaintiffs have not shown an impermissible reduction in effective opportunities for Latino electoral control or in opportunities for Latino participation in the political process.” Ibid.
Viewed against this backdrop, the majority’s holding that Plan 1374C violates §2 amounts to this: A State has denied minority voters equal opportunity to “participate in the political process and to elect representatives of their choice,” 42 U. S. C. § 1973(b), when the districts in the plan a State has created have better prospects for the success of *503minority-preferred candidates than an alternative plan, simply because one of the State’s districts combines different minority communities, which, in any event, are likely to vote as a controlling bloc. It baffles me how this could be vote dilution, let alone how the District Court’s contrary conclusion could be clearly erroneous.
II
The majority arrives at the wrong resolution because it begins its analysis in the wrong place. The majority declares that a Gingles violation is made out “[considering” former District 23 “in isolation,” and chides the State for suggesting that it can. remedy this violation “by creating new District 25 as an offsetting opportunity district.” Ante, at 429. According to the majority, “§2 does not forbid the creation of a noncompact majority-minority district,” but “[t]he noncompact district cannot... remedy a violation elsewhere in the State.” Ante, at 430.
The issue, however, is not whether a § 2 violation in District 23, viewed “in isolation,” can be remedied by the creation of a Latino opportunity district in District 25. When the question is where a fixed number of majority-minority districts should be located, the analysis should never begin by asking whether a Gingles violation can be made out in any one district “in isolation.” In these circumstances, it is always possible to look at one area of minority population “in isolation” and see a “violation” of § 2 under Gingles. For example, if a State drew three districts in a group, with 60% minority voting-age population in the first two, and 40% in the third, the 40% can readily claim that their opportunities are being thwarted because they were not grouped with an additional 20% of minority voters from one of the other districts. But the remaining minority voters in the other districts would have precisely the same claim if minority voters were shifted from their districts to join the 40%. See De Grandy, 512 U. S., at 1015-1016 (“[S]ome dividing by district *504lines and combining within them is virtually inevitable and befalls any population group of substantial size”). That is why the Court has explained that no individual minority voter has a right to be included in a majority-minority district. See Shaw v. Hunt, 517 U. S. 899, 917, and n. 9 (1996) (Shaw II); id., at 947 (Stevens, J., dissenting). Any other approach would leave the State caught between incompatible claims by different groups of minority voters. See Session, supra, at 499 (“[TJhere is neither sufficiently dense and compact population in general nor Hispanic population in particular to support” retaining former District 28 and adding District 25).
The correct inquiry under § 2 is not whether a Gingles violation can be made out with respect to one district “in isolation,” but instead whether line-drawing in the challenged area as a whole dilutes minority voting strength. A proper focus on the district lines in the area as a whole also demonstrates why the majority’s reliance on Bush v. Vera, 517 U. S. 952 (1996), and Shaw II is misplaced.
In those cases, we rejected on the basis of lack of compactness districts that a State defended against equal protection strict scrutiny on the grounds that they were necessary to avoid a § 2 violation. See Vera, supra, at 977-981 (plurality opinion); Shaw II, supra, at 911, 916-918. But those cases never suggested that a plaintiff proceeding under §2 could rely on lack of compactness to prove liability. And the districts in those cases were nothing like District 25 here. To begin with, they incorporated multiple, small, farflung pockets of minority population, and did so by ignoring the boundaries of political subdivisions. Vera, supra, at 987-989 (Appendices A-C to plurality opinion) (depicting districts); Shaw II, supra, at 902-903 (describing districts). Here the District Court found that the long and narrow but more normal shape of District 25 was shared by other districts both in the state plan and the predecessor plan—not to mention the plaintiffs’ own proposed plan—and resulted from the demog*505raphy and geography of south and west Texas. See Session, 298 F. Supp. 2d, at 487-488, 491, and n. 125. And none of the minority voters in the Vera and Shaw II districts could have formed part of a Gingles-compliant district, see Vera, supra, at 979 (plurality opinion) (remarking of one of the districts at issue that it “reaches out to grab small and apparently isolated minority communities which, based on the evidence presented, could not possibly form part of a compact majority-minority district”); Shaw II, 517 U. S., at 916-917 (describing the challenged district as “in no way coincident with the compact Gingles district”); while here no one disputes that at least the Latino voters in the border area of District 25—the larger concentration—must be part of a Latino-majority district if six are to be placed in south and west Texas.
This is not, therefore, a case of the State drawing a majority-minority district “anywhere,” once a §2 violation has been established elsewhere in the State. Id., at 917. The question is instead whether the State has some latitude in deciding where to place the maximum possible number of majority-minority districts, when one of those districts contains a substantial proportion of minority voters who must be in a majority-minority district if the maximum number is to be created at all.
Until today, no court has ever suggested that lack of compactness under §2 might invalidate a district that a State has chosen to create in the first instance. The “geographical!] compact[ness]” of a minority population has previously been only an element of the plaintiff’s case. See Gingles, 478 U. S., at 49-50. That is to say, the § 2 plaintiff bears the burden of demonstrating that “the minority group ... is sufficiently large and geographically compact to constitute a majority in a single-member district.” Id., at 50. Thus compactness, when it has been invoked by lower courts to defeat §2 claims, has been applied to a remedial district a plaintiff proposes. See, e. g., Sensley v. Albritton, 385 F. 3d *506591, 596-597 (CA5 2004); Mallory v. Ohio, 173 F. 3d 377, 382-383 (CA6 1999); Stabler v. County of Thurston, 129 F. 3d 1015, 1025 (CA8 1997). Indeed, the most we have had to say about the compactness aspect of the Gingles inquiry is to profess doubt whether it was met when the district a §2 plaintiff proposed was “oddly shaped.” Growe v. Emison, 507 U. S., at 38, 41. And even then, we rejected §2 liability not because of the odd shape, but because no evidence of majority bloc voting had been submitted. Id., at 41-42.
Far from imposing a freestanding compactness obligation on the States, we have repeatedly emphasized that “States retain broad discretion in drawing districts to comply with the mandate of §2,” Shaw II, supra, at 917, n. 9, and that §2 itself imposes “no per se prohibitions against particular types of districts,” Voinovich v. Quilter, 507 U. S., at 155. We have said that the States retain “flexibility” in complying with voting rights obligations that “federal courts enforcing §2 lack.” Vera, supra, at 978. The majority’s intrusion into line-drawing, under the authority of § 2, when the lines already achieve the maximum possible number of majority-minority opportunity districts, suggests that all this is just so much hollow rhetoric.
The majority finds fault in a “one-way rule whereby plaintiffs must show compactness but States need not,” ante, at 431, without bothering to explain how its contrary rule of equivalence between plaintiffs litigating and the elected representatives of the people legislating comports with our repeated assurances concerning the discretion and flexibility left to the States. Section 2 is, after all, part of the Voting Rights Act, not the Compactness Rights Act. The word “compactness” appears nowhere in §2, nor even in the agreed-upon legislative history. See Gingles, supra, at 36-37. To bestow on compactness such precedence in the §2 inquiry is the antithesis of the totality test that the statute contemplates. De Grandy, 512 U. S., at 1011 (“[T]he ultimate conclusions about equality or inequality of opportunity *507were intended by Congress to be judgments resting on comprehensive, not limited, canvassing of relevant facts”). Suggesting that determinative weight should have been given this one factor contravenes our understanding of how §2 analysis proceeds, see Gingles, 478 U. S., at 45 (quoting statement from the legislative history of §2 that “‘there is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other’”), particularly when the proper standard of review for the District Court’s ultimate judgment under § 2 is clear error, see id., at 78-79.
A § 2 plaintiff has no legally protected interest in compactness, apart from how deviations from it dilute the equal opportunity of minority voters “to elect representatives of their choice.” § 1973(b). And the District Court found that any effect on this opportunity caused by the different “needs and interests” of the Latino voters within District 25 was at least offset by the fact that, despite these differences, they were likely to prefer the same candidates at the polls. This finding was based on the evidence, not assumptions.
Whatever the competing merits of old District 23 and new District 25 at the margins, judging between those two majority-minority districts is surely the responsibility of the legislature, not the courts. See Georgia v. Ashcroft, 539 U. S. 461, 480 (2003). The majority’s squeamishness about the supposed challenge facing a Latino-preferred candidate in District 25—having to appeal to Latino voters near the Rio Grande and those near Austin—is not unlike challenges candidates face around the country all the time, as part of a healthy political process. It is in particular not unlike the challenge faced by a Latino-preferred candidate in the district favored by the majority, former District 23, who must appeal to Latino voters both in San Antonio and in El Paso, 540 miles away. “[Minority voters are not immune from the obligation to pull, haul, and trade to find common political ground, the virtue of which is not to be slighted in applying *508a statute meant to hasten the waning of racism in American politics.” De Grandy, 512 U. S., at 1020. As the Court has explained, “the ultimate right of § 2 is equality of opportunity, not a guarantee of electoral success for minority-preferred candidates of whatever race.” Id., at 1014, n. 11. Holding that such opportunity is denied because a State draws a district with 55% minority citizen voting-age population, rather than keeping one with a similar percentage (but lower turnout) that did not in any event consistently elect minority-preferred candidates, gives an unfamiliar meaning to the word “opportunity.”
Ill
Even if a plaintiff satisfies the Gingles factors, a finding of vote dilution under § 2 does not automatically follow. In De Grandy, we identified another important aspect of the totality inquiry under §2: whether “minority voters form effective voting majorities in a number of districts roughly proportional to the minority voters’ respective shares in the voting-age population.” 512 U. S., at 1000. A finding of proportionality under this standard can defeat § 2 liability even if a clear Gingles violation has been made out. In De Grandy itself, we found that “substantial proportionality” defeated a claim that the district lines at issue “diluted the votes cast by Hispanic voters,” 512 U. S., at 1014-1015, even assuming that the plaintiffs had shown “the possibility of creating more than the existing number of reasonably compact districts with a sufficiently large minority population to elect candidates of its choice,” id., at 1008-1009 (emphasis added).
The District Court determined that south and west Texas was the appropriate geographic frame of reference for analyzing proportionality: “If South and West Texas is the only area in which Gingles is applied and can be met, as Plaintiffs argue, it is also the relevant area for measuring proportionality.” Session, 298 F. Supp. 2d, at 494. As the court ex*509plained, “[ljower courts that have analyzed ‘proportionality’ in the De Grandy sense have been consistent in using the same frame of reference for that factor and for the factors set forth in Gingles.” Id., at 493-494, and n. 131 (citing cases).
In south and west Texas, Latinos constitute 58% of the relevant population and control 85% (six out of seven) of the congressional seats in that region. That includes District 25, because the District Court found, without clear error, that Latino voters in that district “will likely control every primary and general election outcome.” Id., at 504. But even not counting that district as a Latino opportunity district, because of the majority’s misplaced compactness concerns, Latinos in south and west Texas still control congressional seats in a markedly greater proportion—71% (five out of seven)—than their share of the population there. In other words, in the only area in which the Gingles factors can be satisfied, Latino voters enjoy effective political power 46% above their numerical strength, or, even disregarding District 25 as an opportunity district, 24% above their numerical strength. See De Grandy, 512 U. S., at 1017, n. 13. Surely these figures do not suggest a denial of equal opportunity to participate in the political process.
The majority’s only answer is to shift the focus to statewide proportionality. In De Grandy itself, the Court rejected an argument that proportionality should be analyzed on a statewide basis as “flaw[ed],” because “the argument would recast these cases as they come to us, in order to bar consideration of proportionality except on statewide scope, whereas up until now the dilution claims have been litigated on a smaller geographical scale.” Id., at 1021-1022. The same is true here: The plaintiffs’ § 2 claims concern “the impact of the legislative plan on Latino voting strength in South and West Texas,” Session, supra, at 486 (emphasis added), and that is the only area of the State in which they can satisfy the Gingles factors. That is accordingly the proper frame of reference in analyzing proportionality.
*510In any event, at a statewide level, 6 Latino opportunity districts out of 32, or 19% of the seats, would certainly seem to be “roughly proportional” to the Latino 22% share of the population. See De Grandy, supra, at 1000. The District Court accordingly determined that proportionality suggested the lack of vote dilution, even considered on a statewide basis. Session, supra, at 494. The majority avoids that suggestion by disregarding the District Court’s factual finding that District 25 is an effective Latino opportunity district. That is not only improper, for the reasons given, but the majority’s rejection of District 25 as a Latino opportunity district is also flatly inconsistent with its statewide approach to analyzing proportionality. Under the majority’s view, the Latino voters in the northern end of District 25 cannot “count” along with the Latino voters at the southern end to form an effective majority, because they belong to different communities. But Latino voters from everywhere around the State of Texas—even those from areas where the Gingles factors are not satisfied—can “count” for purposes of calculating the proportion against which effective Latino electoral power should be measured. Heads the plaintiffs win; tails the State loses.
* * *
The State has drawn a redistricting plan that provides six of seven congressional districts with an effective majority of Latino voting-age citizens in south and west Texas, and it is not possible to provide more. The majority nonetheless faults the state plan because of the particular mix of Latino voters forming the majority in one of the six districts—a combination of voters from around the Rio Grande and from around Austin, as opposed to what the majority uncritically views as the more monolithic majority assembled (from more farflung communities) in old District 23. This despite the express factual findings, from judges far more familiar with Texas than we are, that the State’s new district would be a *511more effective Latino-majority district than old District 23 ever was, and despite the fact that any plan would necessarily leave some Latino voters outside a Latino-majority district.
Whatever the majority believes it is fighting with its holding, it is not vote dilution on the basis of race or ethnicity. I do not believe it is our role to make judgments about which mixes of minority voters should count for purposes of forming a majority in an electoral district, in the face of factual findings that the district is an effective majority-minority district. It is a sordid business, this divvying us up by race. When a State’s plan already provides the maximum possible number of majority-minority effective opportunity districts, and the minority enjoys effective political power in the area well in excess of its proportion of the population, I would conclude that the courts have no further role to play in rejiggering the district lines under § 2.
I respectfully dissent from Part III of the Court’s opinion.
Justice Scalia, with whom Justice Thomas joins, and with whom The Chief Justice and Justice Alito join as to Part III, concurring in the judgment in part and dissenting in part.
I
As I have previously expressed, claims of unconstitutional partisan gerrymandering do not present a justiciable case or controversy. See Vieth v. Jubelirer, 541 U. S. 267, 271-306 (2004) (plurality opinion). Justice Kennedy’s discussion of appellants’ political-gerrymandering claims ably demonstrates that, yet again, no party or judge has put forth a judicially discernible standard by which to evaluate them. See ante, at 413-423. Unfortunately, the opinion then concludes that appellants have failed to state a claim as to political gerrymandering, without ever articulating what the elements of such a claim consist of. That is not an available disposition of this appeal. We must either conclude *512that the claim is nonjusticiable and dismiss it, or else set forth a standard and measure appellants’ claim against it. Vieth, supra, at 301. Instead, we again dispose of this claim in a way that provides no guidance to lower court judges and perpetuates a cause of action with no discernible content. We should simply dismiss appellants’ claims as nonjusticiable.
II
I would dismiss appellants’ vote-dilution claims premised on § 2 of the Voting Rights Act of 1965 for failure to state a claim, for the reasons set forth in Justice Thomas’s opinion, which I joined, in Holder v. Hall, 512 U. S. 874, 891-946 (1994) (opinion concurring in judgment). As The Chief Justice makes clear, see ante, p. 492 (opinion concurring in part, concurring in judgment in part, and dissenting in part), the Court’s § 2 jurisprudence continues to drift ever further from the Act’s purpose of ensuring minority voters equal electoral opportunities.
III
Because I find no merit in either of the claims addressed by the Court, I must consider appellants’ race-based equal protection claims. The GI Forum appellants focus on the removal of 100,000 residents, most of whom are Latino, from District 23. They assert that this action constituted intentional vote dilution in violation of the Equal Protection Clause. The Jackson appellants contend that the intentional creation of District 25 as a majority-minority district was an impermissible racial gerrymander. The District Court rejected the equal protection challenges to both districts.
A
The GI Forum appellants contend that the Texas Legislature removed a large number of Latino voters living in Webb County from District 23 with the purpose of diminishing Latino electoral power in that district. Congressional redistricting is primarily a responsibility of state legislatures, and legislative motives are often difficult to discern. We pre*513sume, moreover, that legislatures fulfill this responsibility in a constitutional manner. Although a State will almost always be aware of racial demographics when it redistricts, it does not follow from this awareness that the State redistricted on the basis of race. See Miller v. Johnson, 515 U. S. 900, 915-916 (1995). Thus, courts must “exercise extraordinary caution” in concluding that a State has intentionally used race when redistricting. Id., at 916. Nevertheless, when considerations of race predominate, we do not hesitate to apply the strict scrutiny that the Equal Protection Clause requires. See, e. g., Shaw v. Hunt, 517 U. S. 899, 908 (1996) (Shaw II); Miller, supra, at 920.
At the time the legislature redrew Texas’s congressional districts, District 23 was represented by Congressman Henry Bonilla, whose margin of victory and support among Latinos had been steadily eroding. See Session v. Perry, 298 F. Supp. 2d 451, 488-489 (ED Tex. 2004) (per curiam). In the 2002 election, he won with less than 52 percent of the vote, ante, at 423-424 (opinion of the Court), and received only 8 percent of the Latino vote, Session, 298 F. Supp. 2d, at 488. The District Court found that the goal of the map drawers was to adjust the lines of that district to protect the imperiled incumbent: “The record presents undisputed evidence that the Legislature desired to increase the number of Republican votes cast in Congressional District 23 to shore up Bonilla’s base and assist in his reelection.” Ibid. To achieve this goal, the legislature extended the district north to include counties in the central part of the State with residents who voted Republican, adding 100,000 people to the district. Then, to comply with the one-person, one-vote requirement, the legislature took one-half of heavily Democratic Webb County, in the southern part of the district, and included it in the neighboring district. Id., at 488-489.
Appellants acknowledge that the State redrew District 23 at least in part to protect Bonilla. They argue, however, that they assert an intentional vote-dilution claim that is analytically distinct from the racial-gerrymandering claim of *514the sort at issue in Shaw v. Reno, 509 U. S. 630, 642-649 (1993) (Shaw I). A vote-dilution claim focuses on the majority’s intent to harm a minority’s voting power; a Shaw I claim focuses instead on the State’s purposeful classification of individuals by their race, regardless of whether they are helped or hurt. Id., at 651-652 (distinguishing the vote-dilution claim in United Jewish Organizations of Williamsburgh, Inc. v. Carey, 430 U. S. 144 (1977)). In contrast to a Shaw I claim, appellants contend, in a vote-dilution claim the plaintiff need not show that the racially discriminatory motivation predominated, but only that the invidious purpose was a motivating factor. Appellants contrast Easley v. Cromartie, 532 U. S. 234, 241 (2001) (in a racial-gerrymandering claim, “[r]ace must not simply have been a motivation for the drawing of a majority-minority district, but the predominant factor motivating the legislature’s districting decision” (citation and internal quotation marks omitted)), with Arlington Heights v. Metropolitan Housing Development Corp., 429 U. S. 252, 265-266 (1977), and Rogers v. Lodge, 458 U. S. 613, 617 (1982). Whatever the validity of this distinction, on the facts of these cases it is irrelevant. The District Court’s conclusion that the legislature was not racially motivated when it drew the plan as a whole, Session, 298 F. Supp. 2d, at 473, and when it split Webb County, id., at 509, dooms appellants’ intentional-vote-dilution claim.
We review a district court’s factual finding of a legislature’s motivation for clear error. See Easley, supra, at 242. We will not overturn that conclusion unless we are “‘left with the definite and firm conviction that a mistake has been committed.’ ” Anderson v. Bessemer City, 470 U. S. 564, 573 (1985) (quoting United States v. United States Gypsum Co., 333 U. S. 364, 395 (1948)). I cannot say that the District Court clearly erred when it found that “[t]he legislative motivation for the division of Webb County between Congressional District 23 and Congressional District 28 in Plan 1374C was political.” Session, supra, at 509.
*515Appellants contend that the District Court had evidence of the State’s intent to minimize Latino voting power. They note, for instance, that the percentage of Latinos in District 23’s citizen voting-age population decreased significantly as a result of redistricting and that only 8 percent of Latinos had voted for Bonilla in the last election. They also point to testimony indicating that the legislature was conscious that protecting Bonilla would result in the removal of Latinos from the district and was pleased that, even after redistricting, he would represent a district in which a slight majority of voting-age residents was Latino. Of the individuals removed from District 23, 90 percent of those of voting age were Latinos, and 87 percent voted for Democrats in 2002. Id., at 489. The District Court concluded that these individuals were removed because they voted for Democrats and against Bonilla, not because they were Latino. Id., at 473, 508-510. This finding is entirely in accord with our case law, which has recognized that “a jurisdiction may engage in constitutional political gerrymandering, even if it so happens that the most loyal Democrats happen to be black Democrats and even if the State were conscious of that fact.” Hunt v. Cromartie, 526 U. S. 541, 551 (1999). See also Bush v. Vera, 517 U. S. 952, 968 (1996) (plurality opinion) (“If district lines merely correlate with race because they are drawn on the basis of political affiliation, which correlates with race, there is no racial classification to justify”).1 Appellants argue that in evaluating the State’s stated motivation, the District *516Court improperly conflated race and political affiliation by failing to recognize that the individuals moved were not Democrats, they just voted against Bonilla. But the District Court found that the State’s purpose was to protect Bonilla, and not just to create a safe Republican district. The fact that the redistricted residents voted against Bonilla (regardless of how they voted in other races) is entirely consistent with the legislature’s political and nonracial objective.
I cannot find, under the clear error standard, that the District Court was required to reach a different conclusion. See Hunt, supra, at 551. “Discriminatory purpose . . . implies more than intent as volition or intent as awareness of consequences. It implies that the decisionmaker ... selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.” Personnel Administrator of Mass. v. Feeney, 442 U. S. 256, 279 (1979) (citation, some internal quotation marks, and footnote omitted). The District Court cited ample evidence supporting its finding that the State did not remove Latinos from the district because they were Latinos: The new District 23 is more compact than it was under the old plan, see Session, 298 F. Supp. 2d, at 506, the division of Webb County simply followed the interstate highway, id., at 509-510, and the district’s “lines did not make twists, turns, or jumps that can be explained only as efforts to include Hispanics or exclude Anglos, or viceversa,” id., at 511. Although appellants put forth alternative redistricting scenarios that would have protected Bonilla, the District Court noted that these alternatives would not have furthered the legislature’s goal of increasing the number of Republicans elected statewide. Id., at 497. See Miller, 515 U. S., at 915 (“Electoral districting is a most difficult subject for legislatures, and so the States must have discretion to exercise the political judgment necessary to balance competing interests”). Nor is the District Court’s finding at all impugned by the fact that certain legislators *517were pleased that Bonilla would continue to represent a nominally Latino-majority district.
The ultimate inquiry, as in all cases under the Equal Protection Clause, goes to the State’s purpose, not simply to the effect of state action. See Washington v. Davis, 426 U. S. 229, 238-241 (1976). Although it is true that the effect of an action can support an inference of intent, see id., at 242, there is ample evidence here to overcome any such inference and to support the State’s political explanation. The District Court did not commit clear error by accepting it.
B
The District Court’s finding with respect to District 25 is another matter. There, too, the District Court applied the approach set forth in Easley, in which the Court held that race may be a motivation in redistricting as long as it is not the predominant one. 532 U. S., at 241. See also Bush, 517 U. S., at 993 (O’Connor, J., concurring) (“[S]o long as they do not subordinate traditional districting criteria to the use of race for its own sake or as a proxy, States may intentionally create majority-minority districts, and may otherwise take race into consideration, without coming under strict scrutiny”). In my view, however, when a legislature intentionally creates a majority-minority district, race is necessarily its predominant motivation and strict scrutiny is therefore triggered. See id., at 999-1003 (Thomas, J., joined by Scalia, J., concurring in judgment). As in Bush, id., at 1002, the State’s concession here sufficiently establishes that the legislature classified individuals on the basis of their race when it drew District 25: “[T]o avoid retrogression and achieve compliance with §5 of the Voting Rights Act . . . , the Legislature chose to create a new Hispanic-opportunity district—new CD 25—which would allow Hispanics to actually elect its candidate of choice.” Brief for State Appellees 106. The District Court similarly found that “the Legislature clearly intended to create a majority Latino citizen vot*518ing age population district in Congressional District 25.” Session, supra, at 511. Unquestionably, in my view, the drawing of District 25 triggers strict scrutiny.
Texas must therefore show that its use of race was narrowly tailored to further a compelling state interest. See Shaw II, 517 U. S., at 908. Texas asserts that it created District 25 to comply with its obligations under § 5 of the Voting Rights Act. Brief for State Appellees 105-106. That provision forbids a covered jurisdiction to promulgate any “standard, practice, or procedure” unless it “does not have the purpose and will not have the effect of denying or abridging the right to vote on account of race.” 42 U. S. C. § 1973c. The purpose of § 5 is to prevent “retrogression in the position of racial minorities with respect to their effective exercise of the electoral franchise.” Beer v. United States, 425 U. S. 130, 141 (1976). Since its changes to District 23 had reduced Latino voting power in that district, Texas asserts that it needed to create District 25 as a Latino-opportunity district in order to avoid § 5 liability.
We have in the past left undecided whether compliance with federal antidiscrimination laws can be a compelling state interest. See Miller, supra, at 921; Shaw II, supra, at 911. I would hold that compliance with §5 of the Voting Rights Act can be such an interest. We long ago upheld the constitutionality of § 5 as a proper exercise of Congress’s authority trader § 2 of the Fifteenth Amendment to enforce that Amendment’s prohibition on the denial or abridgment of the right to vote. See South Carolina v. Katzenbach, 383 U. S. 301 (1966). If compliance with § 5 were not a compelling state interest, then a State could be placed in the impossible position of having to choose between compliance with § 5 and compliance with the Equal Protection Clause. Moreover, the compelling nature of the State’s interest in § 5 compliance is supported by our recognition in previous cases that race may be used -where necessary to remedy identified past discrimination. See, e. g., Shaw II, supra, at 909 (citing *519Richmond v. J. A. Croson Co., 488 U. S. 469, 498-506 (1989)). Congress enacted § 5 for just that purpose, see Katzenbach, supra, at 309; Beer, supra, at 140-141, and that provision applies only to jurisdictions with a history of official discrimination, see 42 U. S. C. §§ 1973b(b), 1973c; Vera v. Richards, 861 F. Supp. 1304, 1317 (SD Tex. 1994) (recounting that, because of its history of racial discrimination, Texas became a jurisdiction covered by § 5 in 1975). In the proper case, therefore, a covered jurisdiction may have a compelling interest in complying with § 5.
To support its use of § 5 compliance as a compelling interest with respect to a particular redistricting decision, the State must demonstrate that such compliance was its “‘actual purpose’ ” and that it had “ ‘a strong basis in evidence’ for believing,” Shaw II, supra, at 908-909, n. 4 (citations omitted), that the redistricting decision at issue was “reasonably necessary under a constitutional reading and application of” the Act, Miller, 515 U. S., at 921.2 Moreover, in order to tailor the use of race narrowly to its purpose of complying with the Act, a State cannot use racial considerations to achieve results beyond those that are required to comply with the statute. See id., at 926 (rejecting the Department of Justice’s policy that maximization of minority districts was required by §5 and thus that this policy could serve as a compelling state interest). Section 5 forbids a State to take action that would worsen minorities’ electoral opportunities; it does not require action that would improve them.
In determining whether a redistricting decision was reasonably necessary, a court must bear in mind that a State is permitted great flexibility in deciding how to comply with § 5’s mandate. See Georgia v. Ashcroft, 539 U. S, 461, 479-483 (2003). For instance, we have recognized that § 5 does not constrain a State’s choice between creating majority-minority districts or minority-influence districts. Id., at *520480-483. And we have emphasized that, in determining whether a State has impaired a minority’s “effective exercise of the electoral franchise,” a court should look to the totality of the circumstances statewide. These circumstances include the ability of a minority group “to elect a candidate of its choice” or “to participate in the political process,” the positions of legislative leadership held by individuals representing minority districts, and support for the new plan by the representatives previously elected from these districts. Id., at 479-485.
In light of these many factors bearing upon the question whether the State had a strong evidentiary basis for believing that the creation of District 25 was reasonably necessary to comply with § 5,1 would normally remand for the District Court to undertake that “fact-intensive” inquiry. See id., at 484, 490. Appellants concede, however, that the changes made to District 23 “necessitated creating an additional effective Latino district elsewhere, in an attempt to avoid Voting Rights Act liability.” Brief for Appellant Jackson et al. in No. 05-276, p. 44. This is, of course, precisely the State’s position. Brief for State Appellees 105-106. Nor do appellants charge that in creating District 25 the State did more than what was required by §5.3 In light of these concessions, I do not believe a remand is necessary, and I would affirm the judgment of the District Court.
The majority’s fig leaf after stressing the distances involved in District 25—while ignoring the greater ones in former District 23—is to note that “it is the enormous geographical distance separating the Austin and Mexican-border communities, coupled with the disparate needs and interests of these populations—not either factor alone—that renders District 25 noncompact for § 2 purposes.” Ante, at 435. Of course no single factor is determinative because the ultimate question is whether the district is an effective majority-minority opportunity district. There was a trial on that; the District Court found that District 25 was, while former District 23 “did not perform as an effective opportunity district.” Session v. Perry, 298 F. Supp. 2d 451 ,496 (ED Tex. 2004) (per curiam). The majority notes that there was no challenge to or finding on the compactness of *502old District 28, ante, at 435—certainly not compared to District 25—but presumably that was because, as the majority does not dispute, “[ujntil today, no court has ever suggested that lack of compactness under §2 might invalidate a district that a State has chosen to create in the first instance,” infra, at 505. The majority asserts that Latino voters in old District 23 had found an “efficacious political identity,” while doing so would be a challenge for such voters in District 25, ante, at 435, but the latter group has a distinct advantage over the former in this regard: They actually vote to a significantly greater extent. See App. 187 (expert report of R. Gaddie) (for Governor and Senate races in 2002, estimated Latino turnout for District 25 was 46% to 51%, compared to 41.3% and 44% for District 23).
The District Court did not find that the legislature had two motivations in dividing Webb County, one invidious and the other political, and that the political one predominated. Rather, it accepted the State’s explanation that although the individuals moved were largely Latino, they were moved because they voted for Democrats and against Bonilla. For this reason, appellants’ argument that incumbent protection cannot be a compelling state interest is off the mark. The District Court found that incumbent protection, not race, lay behind the redistricting of District 23. Strict scrutiny therefore does not apply, and the existence vel non of a compelling state interest is irrelevant.
No party here raises a constitutional challenge to § 5 as applied in these cases, and I assume its application is consistent with the Constitution.
Appellants argue that in Bush v. Vera, 517 U. S. 952 (1996), we did not allow the purpose of incumbency protection in one district to justify the use of race in a neighboring district. That is not so. What we held in Bush was that the District Court had not clearly erred in concluding that, although the State had political incumbent-protection purposes as well, its use of race predominated. See id., at 969 (plurality opinion). We then applied strict scrutiny, as I do here. But we said nothing more about incumbency protection as part of that analysis. Rather, we rejected the State’s argument that compliance with §5 was a compelling interest because the State had gone beyond mere nonretrogression. Id., at 983; id., at 1003 (Thomas, J., joined by Scalia, J., concurring in judgment).