with whom Justice Souter, Justice Ginsburg, and Justice Breyer join, concurring in the judgment.
The disputed issue of fact in this case is whether political considerations or racial considerations provide the “primary” explanation for the seemingly irregular configuration of North Carolina’s Twelfth Congressional District. The Court concludes that evidence submitted to the District Court on behalf of the State made it inappropriate for that Court to grant appellees’ motion for summary judgment. I agree with that conclusion, but write separately to emphasize the importance of two undisputed matters of fact that are firmly established by the historical record and confirmed by the record in this case.
First, bizarre configuration is the traditional hallmark of the political gerrymander. This obvious proposition is supported by the work product of Elbridge Gerry, by the “swan” designed by New Jersey Republicans in 1982, see Karcher v. Daggett, 462 U. S. 725, 744, 762-763 (1983), and by the Indiana plan reviewed in Davis v. Bandemer, 478 U. S. 109, 183, 185 (1986). As we learned in Gomillion v. Lightfoot, 364 U. S. 339 (1960), a racial gerrymander may have an equally “uncouth” shape. See id., at 340, 348. Thus, the shape of the congressional district at issue in this ease provides strong evidence that either political or racial factors motivated its architects, but sheds no light on the question of which set of factors was more responsible for subordinating any of the State’s “traditional” districting principles.1
*556Second, as the Presidential campaigns conducted by Strom Thurmond in 1948 and by George Wallace in 1968, and the Senate campaigns conducted more recently by Jesse Helms, have demonstrated, a great many registered Democrats in the South do not always vote for Democratic candidates in federal elections. The Congressional Quarterly recently recorded the fact that in North Carolina “Democratic voter registration edges ... no longer translate] into success in statevride or national races. In recent years, conservative white Democrats have gravitated toward Republican candidates.” See Congressional Quarterly Inc., Congressional Districts in the 1990s, p. 549 (1993).2 This voting pattern *557has proved to be particularly pronounced in voting districts that contain more than about one-third African-American residents. See Pildes, The Polities of Race, 108 Harv. L. Rev. 1359, 1382-1386 (1995). There was no need for expert testimony to establish the proposition that “in North Carolina, party registration and party preference do not always correspond.” Ante, at 551.
Indeed, for me the most remarkable feature of the District Court’s erroneous decision is that it relied entirely on data concerning the location of registered Democrats and ignored the more probative evidence of how the people who live near the borders of District 12 actually voted in recent elections. That evidence not only undermines and rebuts the inferences the District Court drew from the party registration data, but also provides strong affirmative evidence that is thoroughly consistent with the sworn testimony of the two members of the state legislature who were most active in drawing the boundaries of District 12. The affidavits of those members, stating that district lines were drawn according to election results, not voter registration, are uncontradicted.3 And almost all of the majority-Democrat registered precincts that the state legislature excluded from District 12 in favor of precincts with higher black populations produced significantly less dependable Democratic results and actually voted for one or more Republicans in recent elections.
The record supports the conclusion that the most loyal Democrats living near the borders of District 12 “happen to be black Democrats,” see ibid., and I have no doubt that the legislature was conscious of that fact when it enacted this apportionment plan. But everyone agrees that that fact is not sufficient to invalidate the district. Cf. ante, at 551-552. That fact would not even be enough, under this Court’s decisions, to invalidate a governmental action, that, unlike the *558action at issue here, actually has an adverse impact on a particular racial group. See, e. g., Personnel Administrator of Mass. v. Feeney, 442 U. S. 256, 279 (1979) (holding that the Equal Protection Clause is implicated only when “a state legislature] 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”); Washington v. Davis, 426 U. S. 229 (1976); Hernandez v. New York, 500 U. S. 352, 375 (1991) (O’Connor, J., concurring in judgment) (“No matter how closely tied or significantly correlated to race the explanation for [a governmental action] may be, the [action] does not implicate the Equal Protection Clause unless it is based on race”).
Accordingly, appellees’ evidence may include nothing more than (i) a bizarre shape, which is equally consistent with either political or racial motivation, (ii) registration data, which are virtually irrelevant when actual voting results were available and which point in a different direction, and (i:i) knowledge of the racial composition of the district. Because we do not have before us the question whether the District Court erred in denying the State’s motion for summary judgment, I need not decide whether that circumstantial evidence even raises an inference of improper motive. It is sufficient at this stage of the proceedings to join in the Court’s judgment of reversal, which I do.
I include the last phrase because the Court has held that a state legislature may make race-based districting decisions so long as those decisions do not subordinate (to some uncertain degree) ‘“traditional... districting principles.”’ See Shaw v. Hunt, 517 U. S. 899, 907 (1996); Miller v. Johnson, 515 U. S. 900, 916 (1995) (holding that racial considerations are subject to strict scrutiny when they subordinate “traditional race-neutral district-ing principles”); id., at 928 (O’Connob, J., concurring) (“To invoke strict *556scrutiny, a plaintiff must show that the State has relied on race in substantial disregard of customary and traditional districting practices”). In this regard, I note that neither the Court’s opinion nor the District Court’s opinion analyzes the question whether the “traditional districting principle” of joining communities of interest is subordinated in the present Twelfth District. A district may lack compactness or contiguity — due, for example, to geographic or demographic reasons — yet still serve the traditional districting goal of joining communities of interest.
The Congressional Quarterly’s the authoritative source regarding the political and demographic makeup of the congressional districts resulting from each decennial census, is even more revealing when one examines its district-by-district analysis of North Carolina’s partisan voting patterns. With regard to the original First District, which was just over 50 percent black, the book remarks: “The white voters of the 1st claim the Democratic roots of their forefathers, but often support GOP candidates at the state and national level. A fair number are ‘Jessecrats,’ conservative Democratic supporters of GOP Sen. Jesse Helms.” Congressional Quarterly, at 550. The book shows that while the Second and Third Districts have “significant Democratic voter registration edges,” Republican candidates actually won substantial victories in four of five recent elections. See id., at 549,552-553. Statistics also demonstrate that a majority of voters in the Eleventh District consistently vote for Republicans “despite a wide Democratic registration advantage.” Id., at 565. Although the book exhaustively analyzes the statistical demographies of each congressional district, listing even the number of cable television subscribers in each district, it does not provide voter registration statistics.
See App. to Juris. Statement 73a (affidavit of Sen. Roy A. Cooper III, Chairman of Senate Redistrieting Committee); id., at 81a-82a (affidavit of Rep. W. Edwin McMahan, Chairman of House Redistrieting Committee).