Justice Kennedy
delivered the opinion of the Court.
The electoral district lines for Georgia’s congressional delegation are before us a second time, appeal now being taken from the trial court’s rulings and determinations after our remand in Miller v. Johnson, 515 U. S. 900 (1995). The three-judge panel of the United States District Court for the Southern District of Georgia was affirmed in Miller after it found the Eleventh Congressional District unconstitutional as then drawn. Race, we held, must not be a predominant factor in drawing the district lines. Id., at 915-917.
Given the contorted shape of the district and the undue predominance of race in drawing its lines, it was unlikely the district could be redrawn without changing most or all of Georgia’s congressional districts, 11 in total number. The [78]*78plan being challenged contained three majority-black districts, and after our remand the complaint was amended to challenge another of these, the then-Second District. The trial court found this district, too, was improperly drawn under the standards we confirmed in Miller. Johnson v. Miller, 922 F. Supp. 1552 (1995).
For the task of drawing a new plan, the court deferred to Georgia’s Legislature, but the legislature could not reach agreement. The court then drew its owm plan, Johnson v. Miller, 922 F. Supp. 1556 (1995); we declined to stay the order; and the 1996 general elections were held under it. The court’s plan contained but one majority-black district. The absence of a second, if not a third, majority-black district has become the principal point of contention. Though the elections have been completed, the plan remains in effect until changed by a valid legislative Act, and the appellants ask us to set it aside.
The private appellants are various voters, defendant-intervenors below, who contend that the interests of Georgia’s black population were not adequately taken into account. The United States, also a defendant-intervenor, joins in the appeal. The state officials, defendants below, do not object to the plan and appeared before us as appellees to defend it. The other set of appellees are the private plaintiffs below, who argued that racial gerrymandering under the previous plan violated their right to equal protection.
The private appellants attack the court’s plan on five grounds. First, citing Upham v. Seamon, 456 U. S. 37 (1982) (per curiam), they say the District Court erred in disregarding the State’s legislative policy choices and in making more changes than necessary to cure constitutional defects in the previous plan. Second and third, they allege the plan violates §§2 and 5 of the Voting Rights Act of 1965, 42 U. S. C. §§ 1973, 1973c. Fourth, they argue the court’s plan contains significant population deviations and so violates the constitutional one-person, one-vote requirement. Fifth, they claim [79]*79the District Court erred in not allowing private intervention on the question of the Second District’s unconstitutionality. The Justice Department included questions one, two, and four in its jurisdictional statement. Private appellants did not brief their fifth contention,- and we will not address it. The remaining challenges are unavailing as well, and we affirm the judgment of the District Court.
H-1
We first address appellants’ argument that the court exceeded the remedial power authorized by our decisions, particularly Upham v. Seamon, supra, by failing to follow policies of the state legislature. When faced with the necessity of drawing district lines by judicial order, a court, as a general rule, should be guided by the legislative policies underlying the existing plan, to the extent those policies do not lead to violations of the Constitution or the Voting Rights Act. 456 U. S., at 43. Much of the argument from the parties centers around what legislative redistricting principles the District Court should have acknowledged in drawing its plan. The appellants say the relevant redistricting guideline should be the three majority-black districts of the pre-cleared plan at issue in Miller v. Johnson; and, if not, the two majority-black districts in an earlier legislative effort. These contentions require us to recite some of the background against which the Georgia Legislature — and later the trial court — attempted to draw the districts.
A
Much of the history is recounted in Miller v. Johnson, and we repeat only some of it here. The need for redistricting arose in 1990 when Georgia, because of its population increase, went from 10 authorized congressional seats to 11. To move ahead with redistricting, a special session of the legislature opened in August 1991. Because Georgia is a covered jurisdiction under § 4(b) of the Voting Rights Act, 42 [80]*80U. S. C. § 1973b(b), § 5 of the Act requires it to obtain either administrative preclearance by the Attorney General or approval by the United States District Court for the District of Columbia for any change in a “standard, practice, or procedure with respect to voting.” 42 U. S. C. § 1973c. The proposed change must not have the purpose or effect “of denying or abridging the right to vote on account of race or color.” Ibid. The legislature submitted a plan to the Attorney General for preclearance on October 1, 1991. See Appendix to this opinion (hereinafter Appendix), fig. 1. The plan contained two majority-black districts, the Fifth and the Eleventh. Previously, Georgia had one majority-black district, the Fifth.
The Department of Justice refused preclearance of this plan in January 1992. It then refused preclearance of a second plan submitted by the legislature, also with two majority-black districts. In its second refusal, the Department of Justice cited several alternative plans proposing three majority-black districts, including one called the “max-black” plan, drafted by the American Civil Liberties Union (ACLU) for the General Assembly’s black caucus. At that point, the General Assembly set out to create three majority-black districts to gain preclearance. See Appendix, fig. 2. The plan as adopted used the ACLU’s max-black plan as a model. One of the three majority-black districts, the Eleventh, was a geographic “‘monstrosity, stretching from Atlanta to Savannah. Its core is the plantation country in the center of the state, lightly populated, but heavily black. It links by narrow corridors the black neighborhoods in Augusta, Savannah and southern DeKalb County.’” 515 U. S., at 909 (quoting M. Barone & G. Ujifusa, Almanac of American Politics 356 (1994)). The district as so drawn served its purpose, however, which was to secure preclear-anee from the Department of Justice.
On November 4, 1992, elections were held under the new plan, and all three majority-black districts elected black can[81]*81didates. In 1994, five white voters from the Eleventh District filed suit in the United States District Court for the Southern District of Georgia, alleging a racial gerrymander in the lines of the Eleventh District, in violation of the Equal Protection Clause as interpreted in Shaw v. Reno, 509 U. S. 630 (1993). The District Court panel found the district invalid, with one judge dissenting. Johnson v. Miller, 864 F. Supp. 1354 (1994).
We affirmed. Miller v. Johnson, 515 U. S. 900 (1995). We rejected appellants’ argument that “regardless of the legislature’s purposes, a plaintiff must demonstrate that a district’s shape is so bizarre that it is unexplainable other than on the basis of race.” Id., at 910. We said “the essence of the equal protection claim recognized in Shaw is that the State has used race as a basis for separating voters into districts.” Id., at 911. And we explained that “[t]he plaintiff’s burden is to show, either through circumstantial evidence of a district’s shape and demographics or more direct evidence going to legislative purpose, that race was the predominant factor motivating the legislature’s decision to place a significant number of voters within or without a particular district.” Id., at 916.
We upheld two principal findings of the District Court indicating race was the predominant factor in constructing the Eleventh District. First, it was “'exceedingly obvious’” from the district’s contorted shape, together with the relevant racial demographics, that it was designed to bring in black populations. Id., at 917 (quoting 864 F. Supp., at 1375). Second, considerable evidence — including the State’s own concessions — showed that the General Assembly was driven by “a predominant, overriding desire” to create three majority-black districts to satisfy the Department of Justice. 515 U. S., at 917. The Justice Department, indeed, “ 'would accept nothing less than abject surrender to its maximization agenda.’” Ibid. (quoting 864 F. Supp., at 1366, n. 11).
[82]*82We then considered whether the race-based districting satisfied strict scrutiny because it was narrowly tailored to achieve a compelling governmental interest. As we noted, “[o]ur presumptive skepticism of all racial classifications” prohibited us “from accepting on its face the Justice Department’s conclusion that racial districting is necessary under the Voting Rights Act.” 515 U. S., at 922. After reviewing the evidence, we concluded that “[ijnstead of grounding its objections on evidence of a discriminatory purpose, it would appear the Government was driven by its policy of maximizing majority-black districts.” Id., at 924.
On remand, the District Court deferred to the Georgia Legislature, giving it time to draw a new congressional map. The Governor called a special session of the General Assembly, which met from August 14 to September 12, 1995. The legislature, however, deadlocked on the congressional reapportionment plan. The Georgia House of Representatives adopted a plan with two majority-black districts, Status Report of Defendants Miller, Howard, and Cleland, Aug. 31, 1995, Record, Pleadings Vol. 11, Doc. No. 295, while the Senate adopted a plan with one, Status Report of Defendants Miller, Howard, and Cleland, Sept. 5, 1995, id., Doc. No. 300. On September 13, 1995, defendants notified the District Court that the legislature was unable to resolve its differences and had adjourned, leaving the District Court to develop a remedy.
Plaintiffs had moved to amend their complaint to challenge the Second District as unconstitutional on the same grounds as the Eleventh District, and the court received additional evidence for the purpose. None of the private defendant-intervenors lived in the Second District and, assuming their lack of standing to defend it, they asked for the addition of other parties. The court disallowed the request, ruling the State could defend this aspect of the plan under review.
The court found that race was the “overriding and predominant factor” in drawing the Second District’s borders. 922 [83]*83F. Supp., at 1553. The district, the court noted, split 12 of the district’s 35 counties, 28 of its precincts, and numerous cities. Linda Meggers, Director of Reapportionment Services for the Georgia General Assembly, was qualified as an expert witness and testified it was not possible to create a majority-black Second District without including the black population centers in Columbus and Muscogee Counties, Albany and Dougherty Counties, and Macon and Bibb Counties, which account for most of these splits. She also testified that in constructing the Second District, she followed the ACLU’s max-black plan. Id., at 1554-1555. As with the Eleventh District, the trial court found no compelling reason for the race-based districting of the Second District sufficient to survive strict scrutiny. The appellants do not appeal the determination by the trial court that the Second District as drawn could not survive scrutiny under the standards set forth in Miller, but they do say the trial court erred in not devising a second majority-black district for its own plan.
During the remedy phase, the defendants proposed a variety of plans. One was the 1991 unprecleared plan passed by the Georgia Legislature, with two majority-black districts. The Eleventh District in the 1991 plan closely resembled the Eleventh District in the precleared plan, which has been found improper. The ACLU submitted four plans. One of these, ACLU 1A, with two majority-black districts, was known as the “least change” plan because it was designed to make the minimal changes perceived to be necessary to correct constitutional defects in the existing plan. Another of the ACLU plans, Abrams A, had three majority-black districts. Abrams A split nine counties in the Second District and three in the Eleventh, and for racial reasons. Yet another plan, Abrams C, had two majority-black districts. And a plan jointly sponsored by John Lewis, a black Democratic Member of the United States House of Representatives from Georgia, and Newt Gingrich, a white Republican [84]*84Member — the Lewis-Gingrich Amici-R plan — contained two majority-black districts. In response, it is said, to a submission by plaintiffs, the Justice Department submitted its “Illustrative Plan.” The Justice Department did not do so, however, until after the evidence closed. The plan contained two majority-black districts and split two counties outside the Atlanta area and numerous precincts. The plaintiffs objected to the submission. The District Court mentioned the Illustrative Plan in its opinion but did not give an explicit ruling on the objection. The late submission prevented the Justice Department’s demographer from being cross-examined about racial motivations, and for this reason its significance must be discounted.
The District Court considered the plans submitted by the various parties and then adopted its own. See Appendix, fig. 3. Noting the Justice Department’s thorough “subversion of the redistricting process” since the 1990 census, it based its plan on the State’s 1972 and 1982 plans. 922 F. Supp., at 1563. The court first had to decide where to locate the new Eleventh District, and did so in an area of significant population growth near Atlanta, so as to displace the fewest counties. It then considered Georgia’s traditional redistricting principles based on maintaining: district cores, four traditional “corner districts” in the corners of the State, political subdivisions such as counties and cities, and an urban majority-black district in the Atlanta area. Protecting incumbents from contests with each other was another factor, which the court subordinated to the others because it was “inherently more political.” Id., at 1565. The District Court stated that, in fashioning a remedy, it considered the possibility of creating a second majority-black district but decided doing so would require it to “subordinate Georgia’s traditional districting policies and consider race predominantly, to the exclusion of both constitutional norms and common sense.” Id., at 1566. Georgia did not have a black population of sufficient concentration to allow creation [85]*85of a second majority-black district, the court found, adding that if it had the court “would have included one since Georgia’s legislature probably would have done so.” Id., at 1567, n. 16. The resulting plan contained one majority-black district, the Fifth. The plan split no counties outside the Atlanta area. The District Court rejected potential objections to the plan based on §§2 and 5 of the Voting Rights Act and the constitutional requirement of one person, one vote.
B
Given this background, appellants say, the District Court’s plan violates our direction in Upham v. Seamon to take account of legislative preferences. In Upham, the District Court considered a reapportionment plan passed by the Texas Legislature. The Attorney General had objected under § 5 of the Voting Rights Act to a specific part of the plan, namely, the lines drawn for two contiguous districts in south Texas. He had approved the other 25 districts. The trial court, required to draw new lines, redrew not just the two districts found objectionable and their neighbors but also some unrelated districts in Dallas County, hundreds of miles to the north. 456 U. S., at 38. In the absence of a finding that the legislature’s reapportionment plan offended either the Constitution or the Voting Rights Act, we held, the District Court “was not free ... to disregard the political program” of the state legislature. Id., at 43. See also White v. Weiser, 412 U. S. 783, 797 (1973).
The instant action presents a quite different situation from Upham, and for several reasons. In the first place, the pre-cleared plan is not owed Upham deference to the extent the plan subordinated traditional districting principles to racial considerations. Upham called on courts to correct — not follow — constitutional defects in districting plans. 456 U. S., at 43. In Miller, we found that when the Georgia Legislature yielded to the Justice Department’s threats, it also adopted the Justice Department’s entirely race-focused ap[86]*86proach to redistricting — the max-black policy. 515 U. S., at 917-918. Using the precleared plan as the basis for a remedy would validate the very maneuvers that were a major cause of the unconstitutional districting.
Second, the constitutional violation here affects a large geographic area of the State; any remedy of necessity must affect almost every district. In Upham, only 2 contiguous districts out of 27 were in violation. Here, as the District Court pointed out, 2 of 11 districts were found unconstitutional, on opposite sides of the State, districts containing between them all or parts of nearly a third of Georgia’s counties. 922 F. Supp., at 1561. Almost every major population center in Georgia was split along racial lines. Under the circumstances, the District Court was justified in making substantial changes to the existing plan consistent with Georgia’s traditional districting principles, and considering race as a factor but not allowing it to predominate. This approach conforms to the rule explained in Upham.
Appellants’ most specific objection under Upham is that the court’s plan does not contain two majority-black districts. In particular, they point to the State’s original 1991 redistricting plan, denied preclearance, which contained two majority-black districts. As we have suggested above, however, the State was subjected to steady Justice Department pressure to create the maximum number of majority-black districts, and there is considerable evidence the State was predominantly driven by this consideration even in developing its 1991 plan. In support of their position, appellants rely on broad assertions in the State’s brief in this Court in Johnson v. Miller that the original plan “was not perceived as a ‘racial gerrymander.’” Brief for Miller Appellants in Miller v. Johnson, O. T. 1994, No. 94-631, p. 49. Against these assertions, appellees point to the testimony of Ms. Meggers, Director of Reapportionment Services for the Georgia General Assembly, that the second majority-black district was originally designed as a concession to the Justice [87]*87Department’s max-black policy. After being presented with a proposed map of the Eleventh District, “[t]he initial response in our office was that’s ridiculous.” “It was said that it doesn’t make any sense and I said maybe not, but... we may get in trouble with the Justice Department if we don’t draw [it] . . . like that and I think that was . . . the main reason” it was originally drawn. Tr. 431-432 (Oct. 30,1995). Ms. Meggers referred to an “understanding” between the leadership in the legislature and the black caucus that a second majority-black district would be created. Id., at 431. The testimony of several legislators indicated that any such understanding was arrived at in the shadow of the Justice Department’s max-black goal, and that all other policies were to give way to this racial consideration. Robert Hanner, chairman of the House Reapportionment Committee, so indicated in his testimony. Id., at 74-75. Sonny Dixon, a member of the House Reapportionment Committee, confirmed this account and said legislators felt pressure from the Justice Department in 1990 to create all possible majority-black districts. Id., at 81. Thomas Murphy, Speaker of the Georgia House of Representatives in 1990 and now, said in his deposition that the initial 1991 reapportionment plan was based on “what we at least perceived to be the direction and instructions of the Justice Department.” Deposition of Thomas B. Murphy, Record 22-23; see also id., at 4, 6. This evidence all refers to development of the original 1991 legislative plan, not the 1992 precleared plan, and thus undermines the contention that the legislature’s original plan should have been controlling on the District Court.
There is strong support, then, for finding the second majority-black district in Georgia’s 1991 unprecleared plan resulted in substantial part from the Justice Department’s policy of creating the maximum number of majority-black districts. It is not Justice Department interference per se that is the concern, but rather the fact that Justice Department pressure led the State to act based on an overriding [88]*88concern with race. Given this background, it would have been most problematic for the trial court to insist on retaining a second majority-black district without regard to other, neutral districting factors. The trial court did not adopt this course. Instead, it gave careful consideration to creation of a second black district on grounds that a black voting population was one factor in drawing a district; and it concluded it could not draw the second majority-black district without allowing that one consideration to predominate over other traditional and neutral districting principles, principles which were a valid expression of legislative policy. There is ample basis in the record to support these conclusions. No other plan demonstrated a second majority-black district could be drawn while satisfying the constitutional requirement that race not predominate over traditional districting principles. The District Court said in its opinion that “[i]f Georgia had a concentrated minority population large enough to create a second majority-minority district without subverting traditional districting principles, the Court would have included one since Georgia’s legislature probably would have done so.” 922 F. Supp., at 1567, n. 16. The statements of several witnesses support the trial court’s independent conclusion it was not possible to do so. Ms. Meggers testified that, unless race was the predominant motive, a second majority-black district could not be drawn in Georgia. Tr. 434-435 (Oct. 30,1995). Speaker Murphy doubted “very seriously” a second majority-black district could be drawn in Georgia without violating the principles we laid down in Miller. Deposition, Oct. 26, 1995, Record 24.
The court found the 1991 unprecleared plan shared many of the constitutional defects of the precleared plan. Among other things, it connected the south DeKalb County urban black population with the mainly rural east Georgian minority population. 922 F. Supp., at 1563, n. 9. Indeed, the Eleventh District in the 1991 plan in many respects was almost the geographical monstrosity it became in the pre-[89]*89cleared plan. The ACLU plans were introduced at the remedial hearing by Selwyn Carter, an employee of the Atlanta-based private Southern Regional Council whose job was to draw and advocate reapportionment plans across the South. Mr. Carter said his “basic goal” in preparing the plans was “[t]o show that it is possible to draw a plan in which African American voters comprise approximately 50 percent of the voting age population of a district and at the same time show that race was not a factor.” Tr. 296 (Oct. 30, 1995). The “least-change” plan, ACLU 1A, has numerous flaws. Besides its high population deviation, to be discussed, the Eleventh District has an iguana-like shape betraying the same invidious purpose we condemned in Miller. The only two plans close to the trial court’s in terms of population deviation are Abrams A and the Justice Department’s Illustrative Plan. Abrams A, with its three majority-black districts, splits nine counties in the Second District and three in the Eleventh, as well as numerous other counties in different parts of the State. The twisted shapes of its Second and Eleventh Districts again bear witness to racial motivation. The Illustrative Plan splits Bibb County — a county never before split in apportionment plans — to subsume Macon’s black population. Although the Justice Department submitted the plan after the close of evidence, and in consequence its demographer could not be cross-examined on the question of racial motivation, the District Court recognized its apparent racial impetus. 922 F. Supp., at 1561, n. 4. Indeed, the Justice Department acknowledged a racial motivation at oral argument before the Court. Tr. of Oral Arg. 12, 16. The Justice Department also suggested it was proper to split Bibb County because the mayor and city council of Macon supported splitting the county and city into different districts. Id., at 13. Macon’s alleged urge to be segregated for congressional districting purposes, however, cannot vitiate the equal protection rights of the Eleventh District’s objecting voters.
[90]*90Interference by the Justice Department, leading the state legislature to act based on an overriding concern with race, disturbed any sound basis to defer to the 1991 unprecleared plan; the unconstitutional predominance of race in the provenance of the Second and Eleventh Districts of the 1992 pre-cleared plan caused them to be improper departure points; and the proposals for either two or three majority-black districts in plans urged upon the trial court in the remedy phase were flawed by evidence of predominant racial motive in their design. In these circumstances, the trial court acted well within its discretion in deciding it could not draw two majority-black districts without itself engaging in racial gerrymandering.
II
The court-ordered plan is not violative of § 2 of the Voting Rights Act. We reject appellants’ contrary position, which is premised on impermissible vote dilution in the court’s failure to create a second majority-black district. Section 2 of the Voting Rights Act applies to any “voting qualification or prerequisite to voting or standard, practice, or procedure ... imposed or applied by any State or political subdivision .....” 42 U. S. C. § 1973(a). On its face, § 2 does not apply to a court-ordered remedial redistricting plan, but we will assume courts should comply with the section when exercising their equitable powers to redistrict. A violation of §2 occurs if “it is shown 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 racial minority]... 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.” 42 U. S. C. § 1973(b).
Our decision in Thornburg v. Gingles, 478 U. S. 30 (1986), set out the basic framework for establishing a vote dilution claim against at-large, multimember districts; we have since extended the framework to single-member districts. Growe [91]*91v. Emison, 507 U. S. 25, 40-41 (1998). Plaintiffs must show three threshold conditions: first, the minority group “is sufficiently large and geographically compact to constitute a majority in a single-member district”; second, the minority group is “politically cohesive”; and third, the majority “votes sufficiently as a bloc to enable it... to defeat the minority’s preferred candidate.” 478 U. S., at 50-51. Once plaintiffs establish these conditions, the court considers whether, “on the totality of circumstances,” minorities have been denied an “equal opportunity” to “participate in the political process and to elect representatives of their choice.” 42 U. S. C. § 1973(b).
The trial court found that to create a second majority-black district in Georgia would require subordinating Georgia’s traditional districting policies and allowing race to predominate. 922 F. Supp., at 1566. We considered the determination in our discussion above and concluded it was well founded. If race is the predominant motive in creating districts, strict scrutiny applies, Bush v. Vera, 517 U. S. 952, 962 (1996), and the districting plan must be narrowly tailored to serve a compelling governmental interest in order to survive. We have assumed, without deciding, that compliance with §2 can be a compelling state interest. See, e. g., id., at 977; Miller v. Johnson, 515 U. S., at 921. Here, there was no “strong basis in evidence,” Shaw v. Reno, 509 U. S., at 656 (internal quotation marks omitted), to conclude that vote dilution, in violation of § 2, would occur in consequence of the court’s plan. In fact, none of the three Gingles factors, the threshold findings for a vote dilution claim, were established here. See Bush, supra, at 976-979.
Here the District Court found, without clear error, that the black population was not sufficiently compact for a second majority-black district. 922 F. Supp., at 1567. So the first of the Gingles factors is not satisfied. As we have noted before, § 2 does not require a State to create, on predominantly racial lines, a district that is not “reasonably [92]*92compact.” Johnson v. De Grandy, 512 U. S. 997, 1008 (1994). And the §2 compactness inquiry should take into account “traditional districting principles such as maintaining communities of interest and traditional boundaries.” Bush, supra, at 977.
The trial court also found the second and third Gingles factors — the extent of racially polarized voting — wanting. In the Eleventh District inquiry, the District Court found that § 2 did not justify drawing racial lines, and it discussed evidence of racial polarization at great length. The court found the statistical evidence was for the most part inconclusive and conflicting, but that the State’s expert, Dr. Joseph Katz, was convincing in his refutation of Dr. Allan Lichtman, the United States’ expert. 864 F. Supp., at 1388. The court found “a significant degree of crossover voting in Georgia and the Eleventh District,” id., at 1390, and that the record “fail[ed] to demonstrate ... chronic bloc voting,” id., at 1392. The court found that the average percentage of whites voting for black candidates across Georgia ranged from 22% to 38%, and the average percentage of blacks voting for white candidates ranged from 20% to 23%. Id., at 1390. As the court noted, “[b]lack and black-preferred candidates in Georgia have achieved many electoral victories in local and statewide elections and have received significant — occasionally overwhelming — support from both black and white voters within the Eleventh Congressional District.” Id., at 1390-1391. The results of the 1992 Democratic primary in the Eleventh District suggested to the court “a general willingness of white voters to vote for black candidates”: black candidates in that primary received about 55% of the white vote, and Cynthia McKinney, a black, won the runoff against a white with 23% of the white vote. Id., at 1391.
For the inquiry concerning the Second District and the remedy, appellants relied exclusively on the Eleventh District trial record. After the remedy hearing, the District Court reaffirmed its earlier findings and cited additional evi[93]*93dence of crossover voting. 922 F. Supp., at 1567. At the hearing concerning the Second District, Ms. Meggers stated that election results in the district indicated significant white crossover voting, and Representative Sanford Bishop, the black congressman elected in the Second District, agreed. Tr. 438, 142 (Oct. 30, 1995).
Appellants take issue with the District Court’s assessment of the level of white crossover voting, but argue that, in any event, the level of polarization the District Court found is sufficient to satisfy the Gingles threshold. Under the circumstances, we cannot say the District Court clearly erred in finding insufficient racial polarization in voting to meet the Gingles' requirements. The results of the 1996 general elections tend to support the District Court’s earlier finding of “a general willingness of white voters to vote for black candidates.” 864 F. Supp., at 1391. All three black incumbents won elections under the court plan, two in majority-white districts running against white candidates. (In Gin-gles, the Court indicated that incumbency is a “special circumstanc[e]” to be taken into account in evaluating racial bloc voting. 478 U. S., at 57. And in this action, the black candidates’ success in two majority-white districts, quite different from their previous districts, is testimony to the “general willingness” of whites to vote for blacks.) These results also underscore the weakness of the Justice Department’s methodology of calculating the likelihood of a black-preferred candidate winning based on strict racial percentages. Brief for United States 27, and n. 18. The Justice Department predicted that a black-preferred candidate “would likely be foreclosed from winning” in the court plan’s Tenth District, and that “[t]he same result would follow even more clearly” in the court’s Fourth District, which had a black voting age population of 33%. Id., at 27. In fact, Representative McKinney won in the Fourth District.
Appellants argue the District Court’s findings on §2 are inconsistent and not owed deference, since the court held § 2 [94]*94required maintenance of the majority-black Fifth District but not creation of a new majority-black district. The District Court found the black population in the Fifth District “is sufficiently compact and, being an urban minority population, has a sufficiently strong community of interest to warrant being a majority-minority district.” 922 F. Supp., at 1568. The court also said the probability of electing a candidate is below 50% when the percentage of black registered voters is 50%, ibid., and therefore the percentage of black registered voters should be kept as close to 55% as possible in the Fifth District. (The District Court noted, however, that it was uncomfortable using percentages of registered voters rather than voting age population, since “that in essence condones voter apathy.” Id., at 1568, n. 18.) The court made no explicit findings about differences in the racial polarization of voting between the Fifth and Eleventh Districts.
We do not agree that the District Court’s maintenance of the Fifth District as a majority-black district under §2 indicates its § 2 findings in reference to other districts are conflicting and not entitled to deference. The District Court noted that maintenance of a majority-black district in the Atlanta area — created in 1972 for compliance with the Voting Rights Act — had become a state districting policy. Id., at 1565. Further, it is possible, although we do not express any opinion on the subject, that changing the racial majority of the district would have violated §5 retrogression principles.
Private appellants also argue no deference is due the District Court’s § 2 finding both because the court did not hold a separate hearing on whether its remedial plan violated § 2 and because it barred private intervention to defend the constitutionality of the Second District. We do not agree. First, neither our precedents nor the Act require the court to hold a separate hearing on the adequacy under §2 of a remedial plan. Second, the private defendant-intervenors [95]*95had ample opportunity to present evidence of the need for a second majority-black district under § 2 at the remedy hearing, in which they fully participated. The finding that appellants have not shown the threshold Gingles factors for a §2 violation is owed deference, and we find it not clearly erroneous.
r-H H-i
The private appellants contend the District Court’s plan also violates §5 of the Voting Rights Act. Although the Justice Department did not include this claim in its jurisdictional statement, it agrees with private appellants and briefed the issue.
As we noted above, § 5 requires covered jurisdictions to obtain either administrative preclearance by the Attorney General or approval from the United States District Court for the District of Columbia for any change in a “standard, practice, or procedure with respect to voting,” and requires that the proposed change “not have the purpose and will not have the effect of denying or abridging the right to vote on account of race or color.” 42 U. S. C. § 1973c. We have explained that “the purpose of § 5 has always been to insure that no voting-procedure changes would be made that would lead to a 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).
The question arises whether a court decree is subject to § 5. We have held that “[a] decree of the United States District Court is not within reach of Section 5 of the Voting Rights Act” such that it must be precleared. Connor v. Johnson, 402 U. S. 690, 691 (1971) (per curiam). The exception applies to judicial plans, devised by the court itself, not to plans submitted to the court by the legislature of a covered jurisdiction in response to a determination of unconstitutionality. McDaniel v. Sanchez, 452 U. S. 130, 148-152 (1981). Here, the District Court made clear it had devised its own plan, a proposition not in dispute. In Sanchez, we [96]*96emphasized language in a Senate Committee Report saying that, although preclearance does not apply to court-devised plans, “‘in fashioning the plan, the court should follow the appropriate Section 5 standards, including the body of administrative and judicial precedents developed in Section 5 cases.’” Id., at 149 (quoting S. Rep. No. 94-295, p. 19 (1975)). This is a reasonable standard, at the very least as an equitable factor to take into account, if not as a statutory mandate.
Appellants, however, have some difficulty fixing on a benchmark against which to measure any retrogression. Private appellants say the benchmark should be either the State’s initial 1991 plan, containing two majority-black districts, or the State’s “policy and goal of creating two majority black districts.” Brief for Appellants 48. The Justice Department, for its part, contends the proper benchmark is the 1992 precleared plan, altered to cure its constitutional defects.
Here, as we have noted above in our discussions of both Upham and § 2, appellants have not demonstrated it was possible to create a second majority-black district within constitutional bounds. So, even were we to accept one of their proposed benchmarks, their desired remedy would be unconstitutional. As it happens, none of appellants’ proposed benchmarks is appropriate. The private appellants’ first proposal was not in effect in Georgia because it was refused preclearance. It thus could not operate as a benchmark under the Attorney General’s regulations:
“In determining whether a submitted change is retrogressive the Attorney General will normally compare the submitted change to the voting practice or procedure in effect at the time of the submission. If the existing practice or procedure upon submission was not in effect on the jurisdiction’s applicable date for coverage . . . and is not otherwise legally enforceable under section 5, it cannot serve as a benchmark, and . . . the [97]*97comparison shall be with the last legally enforceable practice or procedure used by the jurisdiction.” 28 CFR § 51.54(b)(1) (1996).
See also Holder v. Hall, 512 U. S. 874, 883-884 (1994) (“Under § 5, then, the proposed voting practice is measured against the existing voting practice .... The baseline for comparison is present by definition; it is the existing status. . . . [Tjhere is little difficulty in discerning the two voting practices to compare to determine whether retrogression would occur”); Reno v. Bossier Parish School Board, 520 U. S. 471, 478 (1997). There are sound reasons for requiring benchmarks to be plans that have been in effect; otherwise a myriad of benchmarks would be proposed in every case, with attendant confusion. This rule is all the more appropriate when one considers the attempt to use as a benchmark the State’s supposed policy of creating two majority-black districts. And the Justice Department’s proposed benchmark — the 1992 plan shorn of its constitutional defects — was also never in effect. Nor can the 1992 plan, constitutional defects and all, be the benchmark. Section 5 cannot be used to freeze in place the very aspects of a plan found unconstitutional.
The appropriate benchmark is, in fact, what the District Court concluded it would be: the 1982 plan, in effect for a decade. 922 F. Supp., at 1569, n. 20. Appellants have not shown that black voters in any particular district suffered a retrogression in their voting strength under the court plan measured against the 1982 plan. Absent such proof, there is no violation of §5. We reject appellants’ assertion that, even using the 1982 plan as a benchmark, the court’s plan is retrogressive. They claim that under the 1982 plan 1 of the 10 districts (10%) was majority black, while under the District Court’s plan 1 of 11 districts (9%) is majority black, and therefore blacks do not have the same electoral opportunities under the District Court’s plan. Under that logic, each time a State with a majority-minority district was allowed to add [98]*98one new district because of population growth, it would have to be majority-minority. This the Voting Rights Act does not require.
IV
Finally, appellants contend the District Court’s plan violates the constitutional guarantee of one person, one vote under Article I, §2. This provision requires congressional districts to achieve population equality “as nearly as is practicable.” Wesberry v. Sanders, 376 U. S. 1, 7-8 (1964). Court-ordered districts are held to higher standards of population equality than legislative ones. A court-ordered plan should “ordinarily achieve the goal of population equality with little more than de minimis variation.” Chapman v. Meier, 420 U. S. 1, 26-27 (1975); Connor v. Finch, 431 U. S. 407, 414 (1977) (same). Here the District Court was not designing districts to remedy a one-person, one-vote violation, but courts should keep in mind that “absolute population equality [is] the paramount objective.” Karcher v. Daggett, 462 U. S. 725, 732 (1983). Slight deviations are allowed under certain circumstances. Chapman, supra, at 26 (“With a court plan, any deviation from approximate population equality must be supported by enunciation of historically significant state policy or unique features”); Connor, supra, at 419-420 (same); Karcher, supra, at 740 (“Any number of consistently applied legislative policies might justify some variance, including, for instance, making districts compact, respecting municipal boundaries, preserving the cores of prior districts, and avoiding contests between incumbent[s]”).
To help in interpreting what follows, we explain a few terms. Overall population deviation is the difference in population between the two districts with the greatest disparity. Average population deviation is the average of all districts’ deviation from perfect one-person, one-vote allocation. If population allocation in Georgia were perfect, each district would have 588,928 people, according to 1990 census data.
[99]*99Here, the District Court plan has an overall population deviation of 0.35%, and an average deviation of 0.11%. The plan has a lower deviation than: the 1992 plan (with its 0.93% overall deviation and its 0.35% average deviation); the 1982 plan; or “any other plan presented to the Court which was not otherwise constitutionally defective.” 922 F. Supp., at 1561. Private appellants and amici in fact proposed plans with much higher deviations. ACLU 1A, the “least change” plan, had an overall population deviation of 0.94%; Abrams C had an overall deviation of 0.99%; and the Lewis-Gingrich Amici-R plan came in last place with an overall deviation of 1.86%. The only plans with lower overall deviations than the court’s plan were the Justice Department’s Illustrative Plan (0.19%) and the ACLU’s Abrams A (0.29%), whose constitutional infirmities are discussed above.
The District Court recited in detail those state policies and conditions which support the plan’s slight deviations. The court explained Georgia’s “strong historical preference” for not splitting counties outside the Atlanta area, 922 F. Supp., at 1561, and for not splitting precincts, id., at 1562. (The court observed that some splitting of precincts was unavoidable in Cobb County because of noncontiguous annexation patterns, and that it had split some precincts in Clayton County to achieve lower population deviations. Id., at 1562, n. 6.) The court acknowledged that maintaining political subdivisions alone was not enough to justify less than perfect deviation in a court plan. See, e. g., Kirkpatrick v. Preisler, 394 U. S. 526, 533-534 (1969) (“[W]e do not find legally acceptable the argument that variances are justified if they necessarily result from a State’s attempt to avoid fragmenting political subdivisions by drawing congressional district lines along existing county, municipal, or other political subdivision boundaries”). The District Court, in conformance with this standard, considered splitting counties outside the Atlanta area, but found other factors “unique to Georgia” weighed against it. See Chapman, supra, at 26. These in-[100]*100eluded maintaining core districts and communities of interest. Georgia has an unusually high number of counties: 159, the greatest number of any State in the Union apart from the much-larger Texas. These small counties represent communities of interest to a much greater degree than is common, and we agree with the District Court that “such a proliferation” provides “ample building blocks for acceptable voting districts without chopping any of those blocks in half.” 864 F. Supp., at 1377.
In any case, even if we had found the court plan’s population deviation unacceptable, the solution would not be adoption of the constitutionally infirm, because race-based, plans of appellants. Indeed, before this Court at oral argument private appellants acknowledged the remedy for any one-person, one-vote violation would not be creation of a second majority-black district. Tr. of Oral Arg. 28-29. Rather, we would require some very minor changes in the court’s plan— a few shiftings of precincts — to even out districts with the greatest deviations.
That exercise, however, and appellant’s objections to the court plan’s slight population deviations, are increasingly futile. We are now more than six years from the last census, on which appellants’ data is based. The difference between the court plan’s average deviation (0.11%) and the Illustrative Plan’s (0.07%) is 0.04%, which represents 328 people out of a perfect district population of 588,928. The population of Georgia has not stood still. Georgia is one of the fastest-growing States, and continues to undergo population shifts and changes. U. S. Dept. of Commerce, Bureau of Census, Statistical Abstract of the United States 29 (1996) (Table 28) (showing Georgia tied for seventh place among the States in percentage of population growth from 1990 to 1995, with 11.2% growth). In light of these changes, the tinkerings appellants propose would not reflect Georgia’s true population distribution in any event. The Karcher Court, in explaining the absolute equality standard, acknowledged that “census [101]*101data are not perfect,” and that “population counts for particular localities are outdated long before they are completed.” 462 U. S., at 732. Karcher was written only two years from the previous census, however, and we are now more than six years from one. The magnitude of population shifts since the census is far greater here than was likely to be so in Karcher. These equitable considerations disfavor requiring yet another reapportionment to correct the deviation.
V
The task of redistricting is best left to state legislatures, elected by the people and as capable as the courts, if not more so, in balancing the myriad factors and traditions in legitimate districting policies. Here, the legislative process was first distorted and then unable to reach a solution. The District Court was left to embark on a delicate task with limited legislative guidance. The court was careful to take into account traditional state districting factors, and it remained sensitive to the constitutional requirement of equal protection of the laws.
* * *
The judgment of the District Court is affirmed.
It is so ordered.
[Appendix to opinion of the Court follows this page.]
[103]*103[[Image here]]