In the latest chapter of this ongoing litigation we determine whether certain provisions of a 1981 consent decree mandating that the City of Birmingham (City) select employees for promotion based upon their race can withstand scrutiny under Title VII of the Civil Rights Act of 1964 and the Equal Protection Clause of the Fourteenth Amendment. We hold that they cannot.1
I. Background
Appellants are fourteen male, non-black employees of the Birmingham fire rescue service (BFRS) and one male, non-black employee of the City engineering department.2 Appellants assert that the City, acting pursuant to a consent decree it entered in 1981 to resolve then-pending litigation, violated their rights under Title VII and the Equal Protection Clause when it made promotion decisions based upon the race of the person to be promoted. To place the City’s use of race in promoting candidates under the consent decree in context, we first review the pertinent events that led to the consent decree and brought us to this stage of the litigation.
A Events Leading to the Consent Decree
In the mid-1970s, the United States, the Ensley Branch of the NAACP, and seven black individuals (Martin plaintiffs) sued the City and the Personnel Board of Jefferson County (Board) in three separate class actions. The suits charged that the City and the Board had unlawfully discriminated against blacks and women in their hiring and promotion decisions.3 The district court consolidated the three original cases and held two trials. The first trial, in 1976, was held on the limited issue of the validity of the Board’s screening tests for entry-level police and firefighter applicants. The district court found that the tests violated Title VII and ordered the Board to certify a number of black applicants for employment with the City. In re Birmingham Reverse Discrimination Employment Litig., 833 F.2d 1492, 1494 & n. 4 (11th Cir.1987), aff'd sub nom. Martin v. Wilks, 490 U.S. 755, 109 S.Ct. 2180, 104 L.Ed.2d 835 (1989) (BRDEL I). In 1979, the district court held a second trial on the validity of other testing and screening devices used by the Board. Id.
While awaiting the outcome of the second trial, the- parties entered settlement negotiations. Eventually, two consent decrees resulted, one between the plaintiffs and the *1531Board and the other between the plaintiffs and the City. The City’s negotiation process was not without its puzzling aspects. As in any negotiation, the City and the United States exchanged proposals several times. The City initially proposed for itself a fixed quota of 35% across-the-board black appointments to all job openings in the City for five years. The United States, the party pressing the discrimination claim against the City, countered with a proposal that addressed certain job categories individually. With respect to BFRS. promotions, the United States’ response set a lower standard than that initially proposed by the City for promotions of blacks to the lieutenant ranks: it proposed promoting blacks from entry-level firefighter to fire lieutenant at a rate equal to two times the percentage of blacks in the entry-level position.4 That is, because at that time blacks comprised 9% of entry-level firefighters, blacks would receive 18%, not the City’s proposed 35%, of all promotions to fire lieutenant. In response, the City accepted the two-times-representation language, but added for itself a more stringent minimum requirement of 25% black promotions in all job categories, regardless of black representation in the job classification immediately below the promotional job. Under the City’s counter-proposal, then, blacks would receive 25%, rather than the United States’ proposed 18%, of all promotions to fire lieutenant. ■
After additional negotiation and cross-submission of draft decrees, the City’s attorneys presented a settlement proposal to the Birmingham City Council for approval.5 The decree as it read when presented to the City Council set a long-term objective for the City to employ blacks in all jobs within the City in proportion to the representation of blacks in the surrounding Jefferson County labor market, which was 28% at the time. There was no fixed annual percentage for black promotions to fire lieutenant. Instead, the decree presented to the City Council mandated that two of the next four lieutenant promotions would be filled by blacks; thereafter, the City would promote blacks to lieutenant at two times the rate of black representation among firefighters.6 The decree as presented also provided for back pay relief in an unspecified amount to individuals who were the victims of alleged past employment discrimination by the City.7 The City Council *1532passed Resolution 547-81 at its regular meeting the following day, authorizing entry into the consent decree.
Resolution 547-81 notes that the parties had reached substantial agreement on the content of a consent decree designed to end the pending litigation. As we read the resolution, it authorized the City Attorney, with approval of the Mayor, to enter into a decree “embodying such terms” as those that were presented to the Council! Following passage of Resolution 547-81, negotiations continued and produced the final version of the decree, which was subsequently approved by the district court and is still in effect today.8
In the final decree, the City did not admit to “any violation of law, executive order or regulation,” but adopted the following plan to remedy past underrepresentation of blacks and women in City employment:
In order to correct the effects of any un-derrepresentation of blacks and women in the City’s workforce caused by any alleged prior discriminatory employment practices, the City agrees to adopt as a long term goal, subject to the availability of qualified applicants, the employment of blacks and women in each job classification in each department of the City of Birmingham in percentages which approximate their respective percentages in the civilian labor force of Jefferson County as defined by the 1970 Federal Census.
Decree ¶ 5. The final decree also sets a specific annual percentage for black promotions to fire lieutenant different from the one contained in the version presented to the City Council. Instead of setting black fire lieutenant promotions at two times the percentage of black firefighters, the final decree establishes that, each year, 50% of all promotions to lieutenant in the BFRS will be filled by qualified blacks. Decree ¶ 6. Under the decree, for every two lieutenant positions that come open, one must be filled by a black candidate, as long as there are qualified black applicants.9
The decree contains no termination date. Rather, it provides that the district court will retain jurisdiction over the decree and that, after the decree operates for a minimum of six years, any party may move to modify or dissolve it.10
*1533Before giving its final approval to the decree, the district court held a hearing to consider objections of interested parties. BRDEL I, 833 F.2d at 1494. The Birmingham Firefighters Association filed objections and, along with two of its members, moved to intervene as of right in each of the three pending cases, contending that the proposed decree would adversely affect members’ rights. Id. at 1495. The district court denied the motion to intervene and, on August 18,1981, approved the decree. Id. The City has promoted fire lieutenants in the manner prescribed by the final decree since that time. See infra part I.D.
Following denial of the motion to intervene, seven non-black male firefighters sought a preliminary injunction to enjoin operation of the decree, asserting that the promotion provisions would discriminate against them based upon their race, in violation of Title VII. The district court denied the injunction. The district court’s refusal to permit intervention and its denial of the application for a preliminary injunction were consolidated on appeal to this Court. Id. We affirmed the denial of injunctive relief and dismissed the appeal of the denial of intervention, noting that the' complaining firefighters would not be prejudiced by that denial because they could file a separate Title VII action against the City to address whatever harm they might suffer as a' result of the operation of the decree. United States v. Jefferson County, 720 F.2d 1511 (11th Cir.1983).
B. Post-decree History
The non-black male firefighters then brought suit in the district court against the City and the Board, under both the Equal Protection Clause and Title VII; charging reverse discrimination. They asserted that they were being denied promotions because of their race and that the City was promoting allegedly less qualified black firefighters to lieutenant solely on the basis of race, while claiming the City, and Board consent decrees protected its actions from scrutiny. BRDEL I, 833 F.2d at 1495. The City and the Board admitted making “ ‘numerous race conscious promotion and employment decisions pursuant to [the City decree’s] terms.’” Id. at 1496 (alteration in original). They contended that the plaintiffs, although not signatories, were bound by the City and Board decrees and that the promotions in question were lawful because they were made pursuant to the decrees. Id.
In BRDEL I we noted that, whén the non-black firefighters’ claims were tried in 1985, the district court “treated the plaintiffs as if they were bound by the consent decrees and as if they were alleging solely that the City had violated [the terms of] the City decree.” Id. As a result, the district court focused on whether the City had violated paragraph 2 of the decree, which provided that nothing in the decree" required the City to promote a person who was less qualified “ ‘in preference to a person who is demonstrably better qualified based upon the results of a job related selection procedure.’ ” Id. at 1497. The issue tried in 1985 thus was whether the City had complied with the terms of the decree, not the legality of the consent decree itself.
The district court found that the City did not use a job related selection procedure to evaluate candidates referred to it by the Board and that it had made no effort to develop any such procedure. Id. On appeal, we noted that this finding prevented the plaintiffs from establishing a violation of paragraph 2: “since the City did not use a job-related selection procedure, the court apparently reasoned, paragraph 2 imposed no obligations on it.” Id. We held that the district court erred in finding the individual plaintiffs bound by the consent decree and remanded the case for trial on the merits of the claims of unlawful discrimination under Title .VII and the Equal Protection Clause. Id. The Supreme Court affirmed. Martin v. Wilks, 490 U.S. 755, 109 S.Ct. 2180, 104 L.Ed.2d 835 (1989). Thus, in 1991, more than ten years after the City consent decree was entered, the district court held a trial on the Appellants’ claim that the City’s use of race when making promotions in the BFRS *1534violated both Title VII and the Equal Protection Clause.
C. The 1991 Trial
In BRDEL I, we perceived “no reason for treating a consent decree entered pursuant to a voluntary settlement differently from a voluntary affirmative action plan” and rejected “any notion that the memorialization of that voluntary undertaking in the form of a consent decree somehow provides the employer with extra protection against charges of illegal discrimination.” BRDEL I, 833 F.2d at 1501 (footnote omitted).11 We therefore instructed the district court on remand to treat the consent decree as a voluntary affirmative action plan and evaluate it under Title VII, using the analysis articulated in Johnson v. Transportation Agency, 480 U.S. 616, 107 S.Ct. 1442, 94 L.Ed.2d 615 (1987), and under the Equal Protection Clause using the analysis of Wygant v. Jackson Board of Education, 476 U.S. 267, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986). Id. at 1501 & n. 23.
Noting that this case involved race-based employment decisions made by a government employer, the district court recognized that it should apply a strict scrutiny review to the decree. Bennett v. Arrington, 806 F.Supp. 926, 928 (N.D.Ala.1992); see City of Richmond v. J.A. Croson Co., 488 U.S. 469, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989); Peightal v. Metropolitan Dade County, 940 F.2d 1394, 1399 (11th Cir.1991) (“In Croson, a majority of- the Court finally agreed that the constitutionality of a state or local public minority preference program must satisfy a strict scrutiny standard.”), cert. denied, — U.S. -, 112 S.Ct. 969, 117 L.Ed.2d 134 (1992). The district court found that the City was justified in entering the decree because it had a strong basis in evidence for believing that it had discriminated against minorities in the past. Bennett, 806 F.Supp.at 928-29. The district court further found that the City’s use of race under the decree was “limited and tailored to the relief necessary to overcome the employment effects of past discrimination by the City,” while placing an “acceptable burden” on third parties like the Appellants. Id. at 929, 931. Accordingly, it denied the Appellants’ claims. Id: at 931.
The district court did not separately evaluate Appellants’ Title VII and Equal Protection Clause claims. Rather, it analyzed the City decree using the approach announced in Johnson, a Title VII ease, and applied that analysis to all claims raised. We have previously recognized the struggle that district courts face in extracting guidance from the various multi-part Supreme Court opinions on the constitutionality of voluntary, government-sponsored affirmative action programs developed in various contexts. See Peightal, 940 F.2d at 1399 (noting the diversity of views expressed in Supreme Court opinions leading up to Croson ).12 We believe, however, that in the interest of clarity our analysis must separately address Appellants’ Title VII and Equal Protection claims. In order to assist our analysis, we first describe in greater detail the way in which the City uses race to determine candidates for promotion under the consent decree. Our understanding of the operation of the consent decree is informed by the City’s brief on appeal and by *1535the district court’s findings’ of fact' and conclusions of law entered at the close of the first trial on the Appellants’ claims in 1985. See In re Birmingham Reverse Discrimination Employment Litig., 39 Fair Empl. Prac. Cas. (BNA) 1431, 1985 WL 56690 (N.D.Ala.1985).
D. Operation of the City Decree
When a City department has a vacancy in a job classification to be filled by promotion, it sends a “request for certification” of promotional candidates to the Board. Id. at 1438, ¶27. The City’s personnel office reviews the request to determine whether the department is in compliance with the City’s affirmative action plan. Id. If the department is not in compliance, the request is stamped with a notation indicating that the City requests that the Board certify qualified blacks (or females) for the open position. Id.
The Board administers written promotional examinations and tracks employee seniority in City jobs. Id. ¶25. It grades promotional examinations, establishes a'passing score, and calculates a converted score on a standardized scale for persons who pass- an exam. Id. The Board then adds one point to each employee’s converted score for each year of service with the City, up to a maximum of twenty years’ service, to determine a final score. Employees of all races are then ranked on a single eligible register according to their final score. Id.13
The Board follows a “rulé of three” when it supplies the City with names of candidates for promotional openings, providing a number of candidates equal to the number of openings plus two. Id. ¶27.14 That is, if there is one opening, the Board provides three names; if there are four openings, the Board provides six names. If the City indicates on the request for certification that blacks are needed to meet the requirements of the decree, the Board reviews the eligible register and provides the necessary number of names on separate lists of black and non-black eligible candidates. Black candidates and non-black candidates are listed separately in the order in which they are ranked on the eligible register. Id. at 1437-38, ¶¶ 17, 27. For example, if the Board supplies names of four blacks and four non-blacks for promotional openings, the black list contains the four highest-ranked blacks on the eligible register and the non-black list contains the four highest-ranked non-blacks on the eligible register, regardless of how high or low they actually rank on the register relative to each other.
When the black and non-black candidate lists arrive in the BFRS, the Chief reviews the personnel files of the certified individuals and consults with the deputy chief about the candidates to determine whether any candidates are unqualified. Id. at 1437, ¶ 16. The Chief does not compare the relative qualifications of black and non-black candidates for promotion prior to making his selection from the lists.. Nor does he compare the relative qualifications of the individuals certified as eligible against others of the same race certified on the same list. Id. This is because he believes he must promote the highest ranked individual on each list unless he can “prove that the highest ranked individual of either race is unqualified for the promotional position.” Id. at 1437-38, ¶18. Therefore, he typically promotes in the order in which candidates appear on the respective lists. Id. at 1437, ¶16. The Chief alternates selections between the top name on the black list and the top name on the non-black list until all available openings are filled. Id. ¶ 17.15
Under this system, then, employees who will eventually fill an opening in the BFRS are pre-selected by race. If four fire lieutenant positions are open, two of those will be *1536filled by blacks and two by non-blacks. Among employees certified by the Board as eligible candidates, promotions are awarded solely by race. No black employee ever competes for the two openings designated in advance to be filled by non-blacks. No non-black employee can compete for the two openings designated for blacks.
The impact of this system on the Appellants was such that, in the words of the district court:
Each of the plaintiffs who complains in this litigation against the failure to be appointed as a fire lieutenant or fire captain or civil engineer[,] or. who claims that he was delayed in such an appointment[,] was adversely affected because he was white. Those persons[,] in the absence of the consent decree and in the absence of any affirmative action plan adopted by the City as mandated by the decree would, as I interpret the evidence, have been appointed to the positions they desired and about which they here complain. Each of thosé individuals ranked higher on the certification list provided by the Personnel Board than the blacks who were appointed by the City pursuant to the consent decree.
Id. at 1433. Thus, the question we must determine here is whether an affirmative action plan that continuously segregates employees in this manner, and has the above-described impact on the promotional opportunities of employees who are not beneficiaries of its provisions, is permissible under both Title VII and the Equal Protection Clause.
II. Method of Analysis
The Supreme Court established in' Croson that strict scrutiny is the standard of review to be applied to all claims of race-based discrimination by state government entities, regardless of the race of the complaining party. 488 U.S. at 493, 518-21, 109 S.Ct. at 722, 735-36. Accordingly, we apply strict scrutiny to the claims before us. Our analysis of the Appellants’ claims, however, does not begin and end with Croson. Although informing our analysis, Croson is distinguishable from this case in at least two ways. First, Croson did not present a Title VII claim for review. Second, Croson did not involve race-based decisions made by a state governmental entity attempting to remedy its own history of prior employment discrimination.16 Indeed, Johnson, a Title VII case, is more analogous to this case in some respects. Johnson notes that the obligations of a public employer under Title VII and the Constitution are not identical. 480 U.S. at 628 n. 6, 107 S.Ct. at 1449 n. 6 (“The fact that a public employer must also satisfy the Constitution does not negate the fact that the statutory prohibition with which that employer must contend was not intended to extend as far as that of the Constitution.”); id., 480 U.S. at 632, 107 S.Ct. at 1452 (“[W]e do not regard as identical the constraints of Title VII and the Federal Constitution on voluntarily adopted affirmative action plans.”). Because this case presents both Title VII and Equal-Protection Clause claims, we must reconcile.the case law developed in both areas with the direction a divided Supreme Court provided in Croson. To maintain clarity, we separately consider the City decree under both Title VII. and the Equal Protection Clause.
III. Title VII Claim
Title VII reads in pertinent part: (a) It shall be an unlawful. employment practice for an employer—
(1) to fail or refuse to hire ... or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race ...; or
(2) to ... classify his employees ... in any way which would deprive or tend to *1537deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race_
42 U.S.C. § 2000e-2. It is now well established that employers may develop .affirmative action plans designed to “further[ ] Title VII’s purpose of eliminating the effects of discrimination in the workplace.” Johnson, 480 U.S. at 630, 107 S.Ct. at 1451.17 An employer’s decision to account for race in an affirmative action plan is “consistent with Title VII’s objective of break[ing] down old patterns of racial segregation and hierarchy.” Id. at 628, 107 S.Ct. at 1450. Yet, an employer implementing a race-conscious affirmative action plan must assure that race will be considered “consistent with Title VII’s purpose ... and that the interests of those employees not benefiting from the plan will not be unduly infringed.” Id. at 632, 107 S.Ct. at 1452.
Determining whether the affirmative action plan embodied in the City decree was implemented consistent with Title VTI’s purpose and without unduly infringing the interests of Appellants involves a two-part test. We must first determine whether the City’s consideration of the race of promotional candidates was justified by a manifest racial imbalance that reflected under-representation of blacks in traditionally segregated job categories. Id. If such a justification was present when the plan was developed, we must then determine whether the plan itself provides a proper remedy for that imbalance. A remedy is proper if the plan does not unnecessarily trammel the rights of non-black employees or create an absolute bar to their advancement. Id. at 637, 107 S.Ct. at 1455.
A. Justification — Manifest Racial Imbalance ■
When a job requires no special expertise, determining whether a manifest imbalance exists that would justify race-conscious decisionmaking by the employer involves a comparison of the percentage of minority employees in that job with the percentage of minorities in the general area labor market. Id. at 632, 107 S.Ct. at 1452. When a job requires special skills or training, however, the appropriate comparison is to those in the labor market who possess that special skill or training. Id.; see also United States v. City of Miami, 2 F.3d 1497, 1509 (11th Cir.1993) (“It is well established that, in determining whether there is a work force imbalance that justifies affirmative action remedies, the proper comparison is between the minority composition of the work force in question and the qualified minority population in the relevant labor market.”).
This case presents for review those aspects of the City plan relating, to promotions of firefighters to fire lieutenant in the BFRS. Training and experience as a firefighter are special skills required of those who would become fire lieutenants. As such, to determine if an imbalance existed that would justify race-conscious promotion decisions by the City, the appropriate comparison is between black representation in the BFRS lieutenant ranks and black representation among entry-level firefighters. See Stuart v. Roache, 951 F.2d 446, 450-51 (1st Cir.1991) (comparing percentage of minorities in sergeant rank with percentage of minorities in the rank immediately below having years of service necessary to become sergeants), cert. denied, — U.S. -, 112 S.Ct. 1948, 118 L.Ed.2d 553 (1992).18
*1538This comparison is consistent with the comparison used in Johnson, where the Supreme Court addressed a Title VII challenge to an affirmative action plan developed by a county transportation agency (agency). Johnson, a male, brought suit after Joyce, a female, was promoted to road dispatcher instead of him. Joyce and Johnson were among seven applicants certified as eligible for selection after the initial round of applicant interviews. Three agency supervisors conducted a second round of interviews and recommended Johnson for the position. The agency’s affirmative action coordinator, however, acting under a duty to keep the agency’s director informed of opportunities for the agency to accomplish its objectives under its affirmative action plan, recommended to the director that Joyce be promoted. The director heeded the coordinator’s recommendation and selected Joyce over Johnson.
The agency in Johnson had developed an affirmative action plan that authorized it to consider as one factor the sex of a qualified applicant when making promotions within a traditionally segregated job classification where women were significantly under-represented. The agency adopted its plan believing “mere prohibition of discriminatory practices is not enough to remedy the effects of past practices and to permit attainment of an equitable representation of minorities, women and handicapped persons” in the agency’s work force. Johnson, 480 U.S. at 620, 107 S.Ct. at 1446.
The Supreme Court observed without comment that the agency plan adopted as a long-term objective achieving an agency workforce that mirrored in each major job classification the percentage of women in the area labor market. Id. at 633, 107 S.Ct. at 1453. “Even as it did so, however, the Agency acknowledged that such a figure could not by itself necessarily justify taking into account the sex of applicants for positions in all job categories.” Id. Therefore, the agency developed short-term goals for placing women in certain positions, not to be construed “as quotas that must be met, but as reasonable aspirations in correcting the imbalance in the Agency’s work force.” Id. The Court noted that the agency plan did not “dictate mere blind hiring by the numbers [or] hold supervisors to achiévement of a particular percentage of minority employment ... regardless of circumstances such as economic conditions or the number of available qualified minority applicants.”19 Id. Under the plan, women competed directly with all other qualified applicants: “[n]o persons are automatically excluded from consideration; all are able to have their qualifications weighed against those of other applicants.” Id. at 638, 107 S.Ct. at 1455. The plan “posed no danger that personnel decisions would be made by reflexive adherence to a numerical standard.” Id. at 635, 107 S.Ct. at 1454.
The agency plan recognized that women were severely under-represented in the skilled craft job category, the category that included the road dispatcher position Johnson desired. In fact, there were no women among the 238 workers in that category. The Court noted that, with this type of imbalance in a category, and given the agency’s desire to eliminate such imbalances, it' was reasonable to consider Joyce’s sex as one factor among many in the promotion deci*1539sion. Id. at 635, 107 S.Ct. at 1454. Thus, Joyce’s promotion satisfied the first element for Title VII scrutiny because it was made under an affirmative action plan designed to remove imbalances that existed in traditionally segregated job categories. Id. at 637, 107 S.Ct. at 1455.
1. Findings of Manifest Racial Imbalance Required. As a preliminary matter, we must first determine what effect the Court’s opinion in Croson has on Johnson’s requirement that a manifest racial imbalance must exist before a government entity can use race as a criteria in its employment decisions. Appellants assert that, in order for this Court to determine the validity of the decree and avoid the danger that its provisions were simply acts of racial politics rather than proper remedies for past discrimination, the City is required under Croson to have made particularized findings of its own past discrimination prior to entering the decree. Appellants maintain that when the City approved entry into the decree it made no such finding, and they claim that the lack of a formal finding ’of past discrimination by the City causes its decree to fail.
On the surface, Appellants’ position is supported to an extent by the plain language of the decree itself. The decree, in typical settlement agreement language, specifically provides that the City admits to no “violation of law, executive order or regulations,” but goes on to provide a remedy for its “alleged prior discriminatory employment practices.” In essence, the' City, speaking through the decree, refuses to admit that it has discriminated in the past while setting about to remedy the discrimination its adversaries in litigation asserted to have actually occurred. Nonetheless, we believe that in these circumstances the City was not required to make formal findings .about its own past discrimination — it merely had to have a strong basis in evidence, as we demonstrate below.
Our application of the Johnson manifest imbalance test here is'informed by Croson’s discussion of the necessity for a government entity to identify with specificity the discrimination it seeks to remedy through race conscious measures. See generally Croson, 488 U.S. at 492-505, 109 S.Ct. at 721-28. The Supreme Court has recognized that “[ejvidentiary support for the conclusion that remedial action is warranted becomes crucial when the remedial program is challenged in court by nonminority employees.” Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 277, 106 S.Ct. 1842, 1849, 90 L.Ed.2d 260 (1986). There must be sufficient evidence to allow a trial court to “make a factual determination that the employer had a strong basis in evidence for its conclusion that remedial action was necessary.” Id. Unless such a finding is made, an appellate court has no basis upon which to determine whether race-based action was justified as a remedy for prior discrimination by the City. Id.
Appellants’ assertion that Croson requires the City to have made specific findings of its own past discrimination before entering the consent decree ignores one of the important differences between Croson and this case. In Croson, the city was not acting to remedy its own history of discrimination, but was acting to remedy discrimination it claimed had occurred in private industry. The Court found the city council’s justifications for the use of race to be “an amorphous claim that there has been past discrimination in a particular industry [that] cannot justify the use of an unyielding racial quota.” Croson, 488 U.S. at 497, 109 S.Ct. at 724. It makes a great deal of sense to require a city to describe with particularity the findings that led it to conclude that an entire industry had engaged in discrimination before setting about to fashion a race-conscious remedy for that perceived problem. The circumstances of this ease, however, do not compel a similar requirement. See Wygant, 476 U.S. at 289, 106 S.Ct. at 1855 (O’Connor, J., concurring) (“The imposition of a requirement that public employers make findings that they have engaged in illegal discrimination before they engage in affirmative action programs would severely undermine public employers’ incentive to meet voluntarily their civil rights obligations.”).
We do not read Croson to require that a city settling litigation by consent decree declare itself to have violated the law *1540by discriminating in employment, detailing the elements of a potential plaintiffs complaint in the process, before adopting an affirmative action plan designed to remedy the effects of its own past discrimination. See Croson, 488 S.Ct. at 497-505, 109 S.Ct. at 724-28; cf. Cone Corp. v. Hillsborough County, 908 F.2d 908, 913 (11th Cir.) (noting that the Croson plurality intimated that local governments could enact race-conscious remedies “to redress clear instances of discrimination”), cert. denied, 498 U.S. 983, 111 S.Ct. 516, 112 L.Ed.2d 528 (1990). It is, however, a necessity that some finding be made that the City engaged in past discrimination, in order to allow for proper judicial review of the City’s use of race in its affirmative action plan. See Wygant, 476 U.S. at 278 & n. 5, 106 S.Ct. at 1849 & n. 5. (if “race-based state action is taken to remedy prior discrimination by the governmental unit involved ... the very nature of appellate review requires that a factfinder determine whether the employer was justified in instituting a remedial plan”). In this case, the necessary finding of past discrimination was provided by the district court.
In the- opinion from which this appeal arises, the district court found that there was a strong basis in evidence that the City had discriminated against blacks prior-to implementing the affirmative action plan embodied in the consent decree. Bennett, 806 F.Supp. at 929. It noted that blacks were not represented at all in the BFRS lieutenant ranks as a direct result of discriminatory testing for entry-level positions. Id. Further, it noted that there was a significant imbalance between the number of black lieutenants and the number of black firefighters, the jobs from which lieutenants were typically drawn. Id. at 930. At the conclusion of the 1985 trial, the district court found significant evidence of discrimination prior to the time the City entered into the decree. Id. Additional evidence presented in the 1991 trial confirmed those findings. Id. The district court also noted that, at the time the decree was adopted, it had already found certain tests administered by the Board to have an adverse impact on blacks and to be insufficiently job related to be valid under Title VII. Id. at 929. Finally, the district court found that the City entered into the consent decree only when faced with the imminence of another adverse decision by the district court related to “a further trial, with voluminous evidence, attacking scores of other tests and selection devices ... as having a similar adverse impact and insufficient job-relatedness.” Id. As a result, the district court found that the City was justified in using race to remedy its prior discrimination. Id. We conclude that these findings satisfy Cro-son and permit thorough appellate review.
2. Sufficiency of the Findings and Conclusions of Law. We review the district court’s factual findings under the clearly erroneous standard. Newell v. Prudential Ins. Co., 904 F.2d 644, 649 (11th Cir.1990). The district court’s legal conclusions are subject to de novo review. Id.
The district court found that when the City decree was entered in 1981 there were 453 firefighters in the BFRS, 42 of whom were black. Bennett, 806 F.Supp. at 930 n. 6. There were also 94 lieutenants, 31 captains, and 15 battalion chiefs, none of whom were black. Id. Further, the district court had already found that tests administered by the Board for entry-level City jobs in both the police and fire departments were invalid under Title VII because they had a significant adverse impact on. blacks and were not sufficiently job related. Id. at 929 (citing Ensley Branch, NAACP v. Seibels, 616 F.2d 812 (5th Cir.), cert. denied, 449 U.S. 1061, 101 S.Ct. 783, 66 L.Ed.2d 603 (1980)). The district court found that, against this background, coupled with the advice of counsel that the likelihood of victory in the pending litigation was poor, the City was justified in adopting an affirmative action plan. The parties do not dispute these findings, and we do not hold them to be clearly erroneous.
While the Court in Croson criticized the findings made there for only supporting amorphous and generalized claims of societal discrimination, 488 U.S. at 498, 109 S.Ct. at 724, the findings by the district court here demonstrate that the City had strong evidence before it to believe it had engaged in past racial discrimination when it entered the consent decree. The City was aware that *1541the district court had disapproved the hiring tests used by the Board because they had an adverse impact on blacks. The City knew, through counsel, that voluminous evidence attacking other employment tests and selection procedures had been introduced in pending litigation and was advised by counsel that settlement would be a preferred alternative to another adverse decision. The City also knew of the gross statistical disparity between the number of blacks it employed as firefighters and as officers in the BFRS. Given these facts as determined by the district court, the City was justified in implementing an affirmative action plan that provided for the use of race as a remedy for past discrimination. Therefore, the City’s affirmative action plan satisfies the first element of the two-part test established in Johnson. We next consider whether the specific provisions of the City decree provide a proper remedy to address the identified imbalances in traditionally segregated job categories.
B. Remedy—Unnecessarily Trammeling Rights of Non-black Employees
This element of the Johnson test necessarily involves balancing the use of race to secure opportunities for blacks in traditionally segregated jobs against the impact of the plan on non-blacks. As we stated earlier, a proper remedy cannot unnecessarily trammel the rights of non-black employees or create an absolute bar to their advancement. Johnson, 480 U.S. at 637-638, 107 S.Ct. at 1455. On its face, the decree does not create an absolute bar to advancement for non-black employees because it sets aside a fixed percentage of promotions for blacks, leaving the remainder to be filled by all other employees. Our inquiry is thus confined to whether the decree provisions unnecessarily trammel the rights of the non-black employees who do not benefit from the plan.
Unlike the affirmative action plan approved in Johnson, the City decree does not use race as one factor among several that could be considered when making promotion decisions. Rather, the decree makes a Board-certified firefighter’s race the sole factor used to determine eligibility for all fire lieutenant promotions in the BFRS. Promotion-eligible employees are listed separately by race and promotions are made from those separate lists to fill openings allocated by race. The decree, therefore, embodies exactly the sort of “danger that personnel decisions would be made by reflexive adherence to a numerical standard” the Court cautioned against in Johnson. 480 U.S. at 637, 107 S.Ct. at 1454. Unlike the plan approved in Johnson, in which “[n]o persons are automatically excluded from consideration; all are able to have their qualifications weighed against those of other applicants,” id. at 638, 107 S.Ct. at 1455, the City decree specifically excludes all non-black firefighters from consideration for one-half of all fire lieutenant promotions and all black firefighters from consideration for the other half of lieutenant promotions, based solely on their race. And, unlike the plan approved in Johnson, the City’s annual fire lieutenant promotion quota is not a “reasonable aspiration” designed to correct the imbalance in the City’s workforce without unnecessarily trammeling the rights of the Appellants. It is, •instead, a racial quota that must be met whenever openings occur.
There is no precise formula for determining whether an affirmative action plan unnecessarily trammels the rights of non-beneficiaries. When reviewing affirmative action plans involving race based entry-level hiring goals, the.Supreme Court has noted that the impact of the use of race on non-beneficiaries is diffused, spread across all those in society who might desire the entry-level position. See, e.g., Wygant, 476 U.S. at 281, 106 S.Ct. at 1851. Entry-level hiring goals, while burdening some innocent persons, do not impose the same type of injury on non-beneficiaries as that imposed by the use of race to determine employee layoffs. Id. “[L]ayoffs impose the entire burden of achieving racial equality on particular individuals, often resulting in serious disruption of their lives.” Id. at 284, 106 S.Ct. at 1852. The Supreme Court has determined that the burden imposed through race-based layoffs is too intrusive. Id.
This case • involves neither hirings nor layoffs, but instead concerns promotions made under the City’s affirmative action *1542plan. We view the promotion situation as lying somewhere between entry-level hiring and layoffs in terms of the burden permitted on non-beneficiaries. The burden imposed by the City’s plan on non-beneficiaries in the promotion context is not widely diffused like the burden imposed in a hiring situation. It resembles layoffs in that only specific persons are burdened — -those who can reasonably aspire to promotions within the BFRS because they are already employed there and are qualified for promotion. When race is used under the City’s affirmative action plan to determine promotions, it is the non-black employees excluded from consideration for half of all promotion opportunities, and not society in general, who bear the entire burden of the remedy for the City’s past discriminatory behavior. On the other hand, failure to receive a promotion is not as serious a disruption in an employee’s life as that caused by an impermissible race based layoff program. In determining whether the City’s plan unnecessarily trammels the rights of non-black employees, our analysis necessarily recognizes the difference in impact between hiring, layoff, and promotion provisions in affirmative action plans.
Appellees assert that no Supreme Court case dealing with affirmative action hiring and promotion goals has ever found unnecessary trammeling where the plan did not require firings or layoffs of nonminority employees. In effect, Appellees would have us read part two of the Johnson test to say that any affirmative action plan that does not require firings or layoffs of nonminority employees or create an absolute bar to their advancement passes scrutiny under Title VII. We are unwilling to define “unnecessary trammeling” as nothing short of firings and layoffs; an affirmative action plan can unnecessarily trammel the rights of non-minority employees without requiring firings or layoffs. Such is the case with the promotion provisions of the City’s affirmative action plan under review here.
When the plan was created, the City employed 42 black firefighters and 411 non-black firefighters. The decree set aside half of all future promotions to fire lieutenant for those among the 42 black firefighters who became Board-certified as eligible. By virtue of the rigid manner in which the City has used race to determine promotions under the decree, see supra part I.D, 42 black firefighters, because of their race, gained an exclusive claim to half of all fire lieutenant promotions made under the plan, but lost any possibility of competing for the other half. On the other hand, the remaining 411 non-black firefighters saw the number of promotions to which they might aspire, absent the consideration of their race, cut from 100% to half of all lieutenant promotions. The 50% promotion quota was to continue for the duration of the decree, subject to future modification, without regard to the number of lieutenant openings in any given year, the number of black firefighters eligible to become lieutenants at any particular time, or any factor other than race. As the decree operates, no non-black firefighter can even be considered for a promotional job reserved for blacks unless no blacks remain on the Board’s eligible register.
We discern no legitimate basis for the 50% figure ultimately chosen for the annual fire lieutenant promotion quota. The 50% figure selected is completely uninfluenced by the percentage representation of blacks in the firefighter ranks, the feeder job from which lieutenant promotions are filled. The 50% figure appears entirely arbitrary, set at 50% through the settlement bargaining process after the parties failed to agree on the initially proposed 35%, or the subsequently proposed 25%, or on any other arbitrary figure that might have been proposed as the decree was developed by the attorneys for various involved parties.
In our view, the City decree fails under Title VII because the indefinitely-lasting, arbitrarily-selected 50% figure for annual black promotions to fire lieutenant unnecessarily trammels the rights of non-black firefighters by unduly restricting their promotional opportunities through establishment of an arbitrary fixed quota. Had the promotion figure been tied in some reasonable manner to the percentage of blacks in the firefighter ranks, significantly more non-black employees would have received promotions to fire lieutenant. The district court determined in *15431985 that the Appellants, absent the consent decree, “would ... have been appointed to the positions they desired and about which they here complain.” 39 Fair Emp.Prac.Cas. (BNA) at 1433. The- failure to be promoted, caused by the arbitrarily selected 50% promotion figure, no doubt impacted the earnings of Appellants, who are likely heavily dependent on wages for their day-to-day living. See Wygant, 476 U.S. at 281, 106 S.Ct. at 1851. The failure to be promoted, or even a delay in receiving a promotion, also likely impacts earnings long into the future under whatever retirement program the City provides for BFRS employees.
It bears repeating that this case does not involve entry-level affirmative action where the impact of race-based decisionmaking for remedial purposes is minimized as it is spread across society as a whole; we decide nothing about that kind of affirmative action. Because non-black firefighters bear the entire burden of the race-based fire lieutenant promotion remedy in the decree, and because of the immediate and future ramifications of that burden, it is imperative that the remedy be related in some reasonable manner to the representation of blacks among firefighters. We see no such relationship in this case. We recognize that a governmental entity should take immediate steps to remedy its history of discrimination by awarding victims of that discrimination their rightfully earned promotions. See Howard v. McLucas, 871 F.2d 1000, 1003 & n. 5 (11th Cir.) (approving consent decree providing back pay relief and 240 promotions to black civilian Air Force Base employees into “most likely jobs lost by blacks” due to discrimination), cert. denied 493 U.S. 1002, 110 S.Ct. 560, 107 L.Ed.2d 555 (1989). It is not permissible, however, for a government to implement a rigid quota of race-based promotions, to continue indefinitely, with no basis at all for the quota figure selected.
Our review has located no cases approving a state government affirmative action plan where the promotion remedy was not tied in some manner to the representation of minorities in the pool of candidates for promotion. See, e.g., Wilson v. Bailey, 934 F.2d 301, 303 (11th Cir.1991) (approving, in a ease involving some of the same parties present in this litigation, use of race as one factor among many considered when consent decree called for promotion of blacks and females to certain jobs at rates “at least equivalent to their percentage representation in the applicant pool from which such promotions are made”); Davis v. City of San Francisco, 890 F.2d 1438, 1449 (9th Cir.1989) (approving consent decree providing preferential promotions for minorities based on the percentage of minorities in the relevant labor force), cert. denied 498 U.S. 897, 111 S.Ct. 248, 112 L.Ed.2d 206 (1990). Indeed, within the City decree itself there are examples of promotion provisions that are tied to the representation of blacks in the feeder job category. There is no legitimate reason for the annual fire lieutenant promotion figure to be completely unrelated to the population of black firefighters.
Promotional remedies for past discrimination using percentages not tied to the percentage of minorities in the job immediately below the promotion position have withstood scrutiny only when ordered by a court to combat a state government’s steadfast refusal to obey that court’s long-standing order to develop non-diseriminatory promotion procedures. See United States v. Paradise, 480 U.S. 149, 163, 107 S.Ct. 1053, 1063, 94 L.Edüd 203 (1987) (upholding district court order mandating 50% black representation in next group to be promoted to state trooper corporal to provide relief for state’s failure to develop acceptable promotion procedures eleven years after first ordered to do so by the district court). This, however, is not a case like Paradise, where the district court ordered a remedy to combat an egregious refusal.to follow its prior orders to end discriminatory practices. This is a case where, to end pending litigation, the City voluntarily entered into a consent decree containing a quota-based affirmative action plan that cannot withstand scrutiny.
We hold that the fire lieutenant promotion provision of the City’s affirmative action plan violates' Title VII because it unnecessarily trammels the rights of non-black firefighters by establishing a rigid, arbitrarily selected quota of 50% annual black promotions to fire lieutenant.
*1544IV. Equal Protection Clause Claim
We turn next to Appellants’ Equal Protection Clause challenge to the City’s affirmative action plan. Appellants assert that they were denied promotions to fire lieutenant because of their race in violation of the Equal Protection Clause.20 Appellees maintain that the use of race in determining fire lieutenant promotions should survive an equal protection challenge because the City had a compelling interest in adopting the decree and the provisions of the decree are narrowly tailored to achieve that interest.
“ ‘Racial and ethnic distinctions of any sort are inherently suspect and thus call for the most exacting judicial examination.’ ” Wygant, 476 U.S. at 273, 106 S.Ct. at 1846 (quoting Bakke, 438 U.S. at 291, 98 S.Ct. at 2748). This strict scrutiny is employed even though the racial classification challenged operates against a group not historically subject to discrimination by government.. Id.; see also. Croson, 488 U.S. at 494, 109 S.Ct. at 722 (“We thus reaffirm the view ... that the standard of review under the Equal Protection Clause is not dependent on the race of those burdened or benefited by a particular classification.”). To assure a proper balance between the individual rights secured by the Fourteenth Amendment and the permissible burden placed on innocent persons, the use of race in government decisionmaking is subject to strict judicial scrutiny “to ‘smoke out’ illegitimate uses of race by assuring that the legislative body is pursuing a goal important enough to warrant use of a highly suspect tool.” Croson, 488 U.S. at 493, 109 S.Ct. at 721.
To survive a challenge under the Equal Protection Clause, the City’s use of race in making fire lieutenant promotion decisions must be justified by a compelling interest and must be narrowly tailored to achieve that interest. See Wygant, 476 U.S. at 274, 106 S.Ct. at 1847; Cone Corp., 908 F.2d at 913 (“Under strict scrutiny analysis, the racial classifications must be necessary and must be narrowly tailored to achieve the goal of remedying the effects of past discrimination.”). The City must show that a compelling governmental purpose justified the use of a racial preference and that the means chosen — in this ease, the selection of candidates for promotions solely because of their race — was narrowly tailored to achieve that purpose. See Hayes v. North State Law Enforcement Officer Ass’n, 10 F.3d 207, 212-13 (4th Cir.1993). For the reasons set forth below, we hold that the City had a compelling interest in providing a remedy for its prior discriminatory employment practices, but that the City’s use of race in its affirmative action plan is not narrowly tailored to achieve that interest as required under the Equal Protection Clause.
A.' Compelling Government Interest
Public employers operate under a clear command from the Supreme Court to eliminate the vestiges of prior racial segregation and discrimination. Wygant, 476 U.S. at 276, 106 S.Ct. at 1848. At the same time, public employers must “‘do away with all governmentally imposed discriminations based on race.’” Id. (quoting Patmore v. Sidoti, 466 U.S. 429, 430-33, 104 S.Ct. 1879, 1881-82, 80 L.Ed.2d 421 (1984)). The Supreme Court has noted that these two commands must be reconciled with great care, because they are not always harmonious. Id-That is, while attempting to eliminate the vestiges of past discrimination against persons of one race, government must minimize the imposition of new discriminations on persons of other races. Therefore, the Court requires a showing that the governmental unit involved engaged in prior discrimination, and requires a strong basis in evidence that remedial action is warranted, before allowing that governmental unit to execute an affirmative action plan remedying prior discrimination by using race in a narrowly tailored manner. Croson, 488 U.S. at 499, 109 S.Ct. at 725.
As an initial matter, then, we must decide whether there was a strong basis in evidence that remedial action was warranted *1545before the City entered the consent decree embodying its affirmative action plan.21 As noted previously, the district court, found strong evidence that the City had engaged in prior discrimination in the BFRS. See supra part III.A.2. We accept the district court’s findings and will not repeat them here. We also accept the district court’s findings and conclusion that action to remedy the effects of prior discrimination in the BFRS was warranted. But even when a city is justified in implementing an affirmative action plan, only a plan that is “carefully constructed” will do. See Wygant, 476 U.S. at 286, 106 S.Ct. at 1853 (O’Connor, J., concurring) (“The Court is in agreement thát ... remedying past or present racial discrimination by a state actor is a sufficiently weighty state interest to warrant the remedial use of a carefully constructed affirmative action program.”).
B. , Narrowly Tailored Means
Classifications based upon race carry a very real danger of harm because they “threaten to stigmatize individuals by reason of their membership in a racial group." Shaw v. Reno, - U.S. , 113 S.Ct. 2816, 2824, 125 L.Ed.2d 511 (1993). "{E]ven in the pursuit of remedial objectives, an explicit policy of assignment by race may serve to stimulate our society's latent race consciousness, suggesting the utility and propriety of basing decisions on a factor that ideally bears no relationship to an individual's worth or needs." United Jewish Orgs., Inc. v. Carey, 430 U.S. 144, 173, 97 S.Ct. 996, 1014, 51 L.Ed.2d 229 (1977) (Brennan, J., èoncurring in part and dissenting in part). Preferences based on race "may only reinforce common stereotypes holding that cer-tam groups are unable to achieve success without special protection based on a factor having no relationship to individual worth." Bakke, 438 U.S. at 298, 98 S.Ct. at 2752. As a result, a race conscious government policy justified by a compelling purpose-remedying past racial discrimination-must also use race hi as limited a manner as possible to accomplish that compelling purpose.
Several factors determine whether race-based promotional relief is narrowly tailored to accomplish a compelling purpose, including: “the necessity, for the relief and the efficacy of alternative remedies; the flexibility and duration of the relief, including the availability of waiver provisions, the relationship of numerical goals to the relevant labor market, and the impact of the relief on the rights of [the Appellants].” Howard v. McLucas, 871 F.2d 1000, 1008 (11th Cir.) (internal quotations omitted), cert. denied, 493 U.S. 1002, 110 S.Ct. 560, 107 L.Ed.2d 555 (1989). We address each of these in turn.
1. Necessity for the Relief Granted and Efficacy of Alternative Remedies. In Howard, we explored in detail the provisions of a consent decree using race to provide relief to victims of past discrimination by a government employer. The consent decree provided promotions for.240 specified plaintiffs who likely were previously denied , promotions because of their race. The decree identified 38 target positions, based upon historical career progression patterns, to determine the jobs most likely lost by the plaintiffs and limited the race-conscious promotions to those positions. Id. at 1003. The 240 discrimination victims were listed on special promotion registers and alternated with persons on a general promotion register to fill openings as they occurred in each of the 38 target positions. Id. at 1008. In reviewing a challenge to the decree, we rejected the argument that promotional relief was unnecessary because other alternatives were available. Id. at 1009. We found that the proposed alternatives were not feasible because they did not place the plaintiffs, who were identified victims of prior discrimination, in their rightful place as expeditiously as did the relief contained in the decree. Id.
Where relief is not provided to identified victims of discrimination, given the odious nature of race-based decisionmaking, race-neutral alternatives should be considered before a. government implements an *1546affirmative action plan using race as the sole criteria upon which promotions are based. See Croson, 488 U.S. at 507, 109 S.Ct. at 729 (faulting the city for failing to consider the use of race-neutral means to increase minority business,participation in city contracting); Paradise, 480 U.S. at 171, 107 S.Ct. at 1066 (describing the need to consider the efficacy of alternative remedies when determining whether race-conscious remedies are appropriate). In examining this requirement in this case, the district court found that the City “did attempt to correct its past discrimination in or around 1974, prior to the entry of the consent decree” in 1981. Bennett, 806 F.Supp. at 929. The district court noted that an affirmative action ordinance designed to increase minority hiring by the City was enacted by the City Council in 1974, but was later vetoed by the Mayor. Id. at 929-30. The district court also noted that a subsequent measure was enacted which placed responsibility on various department heads to set and achieve minority employment goals. Id. at 930. The district court found that “[t]hese alternative measures were not effective, as evidenced by the hiring of only two blacks in the Fire Department by 1974” and by the fact that, as of the month before the consent decree was formally entered in 1981, “only 9.3% of the firefighters were black.” Id.
The district court’s finding that alternative measures to the race-based promotion plan in the City decree were not effective is clearly erroneous. First, the fact that only two blacks were hired in the BFRS by 1974 has no bearing on the effectiveness of affirmative action efforts implemented after 1974. Second, post-1974 hiring data indicate that the City had made significant progress in adding blacks to the firefighter ranks prior to entering the consent decree.22 The effectiveness of the City’s pre-decree efforts in hiring blacks as firefighters appears in the chart below:
[[Image here]]
As-the chart demonstrates, between 1978 and 1981, the City increased the number of black firefighters from eight to forty-two — a five-fold increase that was achieved in the absence of the race-based affirmative action plan embodied in the consent decree. We regard this progress as encouraging, not ineffective. Unlike .the situation in Croson, where there was no evidence in the record that the city had even considered alternatives to race-based quotas, Croson, 488 U.S. at 507, 109 S.Ct. at 729, there is strong evidence in the record of this ease that the City had implemented effective alternatives to race-based quotas to remedy its prior discriminatory behavior. While the district court correctly concluded that, when the decree was entered, no black had as yet become a fire lieutenant, we believe that, given the City’s progress at the entry-level, alternative measures designed to increase black representation in the fire lieutenant ranks were feasible.
Appellants suggest several feasible alternatives to the race-based promotion remedy implemented by the City that could have made a significant difference in the representation of blacks among fire lieutenants. For *1547example, Appellants propose eliminating the addition of seniority points to the test score when determining the final score that dictates an employees’ place on the eligible register. See supra part I.D. This race-neutral alternative would prevent non-black firefighters from benefiting in the promotion context from the City’s prior discrimination against blacks in entry-level hiring. The non-black status quo in the fire lieutenant ranks would not be preserved as a result of a seniority system that is a vestige of the City’s prior discrimination against blacks. Under such a system, recently-hired blacks who score higher on promotion exams than more senior non-blacks would rank higher on the eligibility register and be promoted sooner than their lower ranking but more senior non-black co-workers. Appellants also suggest that, if the Board’s promotion exams had an adverse impact on blacks, a race-neutral alternative to those exams would be implementation of another test that has no such adverse impact. In addition, we note that the City could have implemented an affirmative action plan, like the one approved in Johnson, that accounted for the race of a candidate as one factor among many to be considered. The record indicates that, in spite of its progress at the entry level, the City considered no feasible alternatives to the race-based promotional quota system for fire lieutenants before it implemented this system as a remedy for prior discrimination. Considering the efficacy of alternative remedies, the relief provided in the decree cannot be reconciled with the requirement that a government’s use of race must be narrowly tailored.
2. Flexibility and Duration of Race-Based Relief. In Howard, we identified several aspects of the consent decree that readily demonstrated its flexibility, short duration, and minimal adverse effects on non-beneficiaries. Howard, 871 F.2d at 1009. First, the Howard decree did not prevent non-black employees from being promoted because promotions to the target positions from the special promotion roster alternated with those from the general roster, while leaving most positions to be filled by open competition. Id. Second, the plaintiffs had to meet certain qualification criteria in order to be promoted. Id. Third, the government was not required to make unnecessary promotions .simply for- the sake of providing relief to the plaintiffs. Id. Fourth, the relief was not intended- “to set employment percentage goals or ensure a racially balanced workforce, and it evaporate[d] when the 240 promotions [were] made.” Id. A comparison of the decree we approved in Howard with the City decree in this case reveals some similarities, but some more significant differences.
Like the Howard decree, the City decree does not require unnecessary promotions in providing relief. The City decree also requires that persons who benefit from the race-based promotional relief must bé Board-certified as eligible for promotion. Further, the City decree does not prevent non-black firefighters from being promoted to fire lieutenant, calling instead for alternating promotions between blacks and non-blacks. In Howard, however, the burden of the promotion plan on non-black employees was more diffused than the burden on non-black firefighters in this case. There were numerous promotion opportunities in Howard for which black and non-black employees might compete, irrespective of race, beyond the 38 positions targeted by the decree. By comparison, in this case the only promotional position for which a firefighter might reasonably compete is fire lieutenant. A firefighter would be unlikely to compete for promotions outside the career path of the BFRS. Therefore, although the City decree does not prevent promotions of non-black employees, it places a greater burden on a smaller number of non-black firefighters than the Heyw-ard decree placed on larger numbers of non-black Air Force Base employees.
There is absolutely no similarity, however, between the decree approved in Howard and the City decree in the overall objective and operation of the two decrees. The Howard decree was intended to provide specific relief to victims of past discrimination. It did not attempt to ensure a .racially balanced workforce through “blind hiring by the numbers ... amount[ing] to a rigid and impermissible quota system.” Id. The overall objective of the City decree, on the other hand, is for the City to employ blacks in each job classifica*1548tion in each of its departments in percentages which approximate the percentage of blacks in the civilian labor force of the surrounding county. Decree ¶5.
To achieve that objective among fire lieutenants, the City decree sets a 50% annual quota for promotions of blacks. As we explained earlier, in order to meet both its overall objective and its annual quota, the City classifies firefighters by race and makes promotion decisions based solely upon the race of the candidate. For every two promotions to fire lieutenant, one is made from the list of eligible black firefighters and one is made from the list of eligible non-black firefighters. Such use of race by the City is not narrowly tailored to provide a proper remedy for past discrimination. Rather, it is designed to achieve the unconstitutional objective of outright racial balancing.
The Supreme Court has determined that it is “ ‘completely unrealistic’ [to assume] that minorities will choose a particular trade in lockstep proportion to their representation in the local population.” Croson, 488 U.S. at 507, 109 S.Ct. at 729; see also Sheet Metal Workers’ Int’l Ass’n v. EEOC, 478 U.S. 421, 493, 106 S.Ct. 3019, 3059, 92 L.Ed.2d 344 (1986) (O’Connor, J., concurring in part and dissenting in part) (“[I]t is completely unrealistic to assume that individuals of each race will gravitate with mathematical exactitude to each employer or union absent unlawful discrimination.”). This determination comes despite the Court’s acknowledgment that “absent explanation, it is ordinarily to be expected that nondiscriminatory hiring practices will in time result in a work force more or less representative of the racial and ethnic composition of the population in the community from which employees are hired.” International Bhd. of Teamsters v. United States, 431 U.S. 324, 340 n. 20, 97 S.Ct. 1843, 1856 n. 20, 52 L.Ed.2d 396 (1977). There is, however, a considerable difference between expecting to find a balanced work force, absent discrimination, and a governmentally imposed plan using race to ensure a racially balanced work force. See id. at 340 n. 20, 1856 & n. 20 (noting that, while statistics showing racial imbalance are a probative evi-dentiary tool often demonstrating telltale signs of racial discrimination, they cannot support “an erroneous theory that Title VII requires an employer’s work force to be racially balanced”).
The City’s use of the 50% annual quota to establish a work force of fire lieutenants reflecting black employment in lockstep proportion to the proportion of blacks in the local labor force cannot withstand strict scrutiny. See Croson, 488 U.S. at 507, 109 S.Ct. at 729 (finding that a city’s 30% contract set-aside for minority businesses “cannot be said to be narrowly tailored to any goal, except perhaps outright racial balancing”); Cone Corp., 908 F.2d at 914 (defining narrow tailoring in the context of a minority business enterprise plan to “mean[] that the plan must be designed to further some goal other than outright racial balancing”); Mann v. City of Albany, 883 F.2d 999, 1005 (11th Cir.1989) (noting that a majority of the Supreme Court has “expressed clear hostility toward government affirmative-action programs that distribute opportunities woodenly and inflexibly on the basis of an applicant’s race”). The City decree institutionalizes race as the sole criteria by which the City selects certified candidates for promotion to fire lieutenant.
The City’s rigid approach, while administratively convenient, is not a narrowly tailored means to remedy prior discrimination. It is instead an approach designed to achieve government-mandated racial balancing — the perpetuation of discrimination by government. We can imagine nothing less conducive to eliminating the vestiges of past discrimination than a government separating its employees into two categories, black and non-black, and allocating a rigid, inflexible number of promotions to each group, year in and year out. We conclude that the City’s use of race in its affirmative action plan is not narrowly tailored to achieve a compelling government interest. Instead, it provides for “blind hiring by the numbers amount[ing] to a rigid and impermissible quota system.” Howard, 871 F.2d at 1009.
3. Relationship of Numerical Goals to the Relevant Labor Market. We explained above that fire lieutenants are promoted from the ranks of firefighters, making *1549firefighters the relevant comparison group to which the numerical goals of the decree must be related. See supra part III.A. We have also determined that there is no relationship between the numerical goal for fire lieutenant promotions and the representation of blacks among firefighters. As such, the City’s affirmative action plan does not. use race in a narrowly tailored manner.
4. Impact of the Relief on Appellants. The relief provided by the decree in Howard had a limited impact on non-black employees. Certain promotions of non-black employees were delayed while the victims of past discrimination were afforded relief. This impact was limited to the length of time it took to achieve the legitimate objective of providing promotional relief to discrimination victims and the decree evaporated after the 240 promotions were made. Howard, 871 F.2d at 1009-10. In addition, non-black employees remained eligible for and experienced no delay in pursuing promotions to any of the multitude of jobs outside the 38 targeted by the decree. In fact, in the twenty-month period during which the decree operated, promotions made pursuant to the decree amounted only to slightly more than 4% of all promotions made, Id. at 1010. Under those circumstances, we determined that the burden of the race-based promotion decisions imposed on non-black employees did not “fall upon a narrow segment of the work force” and was “relatively diffuse.” Id. at 1009-10.
The impact of the City decree on the non-black Appellants in' this case is much more focused than that of the Howard decree. First, the alternating promotion aspect of the Howard decree delayed some promotions only until the legitimate objective of providing relief to identified discrimination victims was met. The provisions of the City decree, on the other hand, continue in place until the City achieves its unconstitutional racial balancing objective or until the district court terminates the decree. Second, while the non-black intervenors in Howard had open to them many other promotion opportunities beyond those affected by the decree, in this case non-black firefighters with more than two years’ experience could only reasonably expect to compete for promotions, in the BFRS to fire lieutenant. When the City decree was implemented, those among the 42 black firefighters who had two years’ experience found half of all promotions to lieutenant immediately reserved for competition between them and only them, as a result of their race. Those among the 411 non-black firefighters with two years’ experience were consigned by their races to competing for the other half of promotions to lieutenant. Thus, the impact of the decree on non-black firefighters in this case was not “relatively diffuse,” but was instead quite pointed.
In Howard, we found that race was used by the government to provide affirmative remedies for prior discrimination in as narrow a manner as possible under the circumstances. In this ease, however, as we concluded above, the City did not use race as a narrowly tailored remedy for prior discrimination. “Under strict scrutiny the means chosen to accomplish the State’s asserted purpose must be specifically and narrowly framed to accomplish that purpose.” Wygant, 476 U.S. at 280, 106 S.Ct. at 1850. The use of race by government in its decision-making is “simply too pernicious to permit any but the most exact connection between justification and classification.” Fullilove v. Klutznick, 448 U.S. 448, 536, 100 S.Ct. 2758, 2805, 65 L.Ed.2d 902 (1980) (Stevens, J., dissenting). The City decree uses race as an overbroad remedy, categorizing BFRS employees by race and promoting them woodenly and reflexively in rigid adherence to a quota system. This broad brush approach, designed to achieve outright racial balancing through an inflexible but administratively convenient quota system, and implemented without consideration of feasible, race-neutral alternatives designed to remedy prior discrimination, is unconstitutional. The decree violates the Appellants’ rights under the Equal Protection Clause of the Fourteenth Amendment.
Y. Conclusion
We hold that the fire lieutenant promotion provisions of the affirmative action plan embodied in the 1981 City decree violate both Title VII of the Civil Rights Act of 1964 and the Equal Protection Clause of the Four*1550teenth Amendment. We affirm the district court’s holding that the Civil Rights Act of 1991 does not bar Appellants’ claims. On remand, the district court should provide appropriate relief consistent with this opinion. Accordingly, the judgment of the district court is
AFFIRMED in part, REVERSED in part, and REMANDED.
. We affirm without discussion the district court’s holding that § 108 of the Civil Rights Act of 1991 is inapplicable because the conditions for application of that section, even if retroactive application is appropriate, are not present in this case.
. The engineering department employee has been included among the Appellants since the original trial of Appellants’ case in 1985. No material aspect of the claims relating to the engineering department is discussed in the district court opinion that is the subject of this appeal. We will, therefore, confine our analysis to the claims raised regarding the BFRS. On remand, the district court should apply our analysis and holdings to all Plaintiffs with pending claims.
.The Board and the City share responsibility for hiring and promoting City employees. The Board administers tests, tracks employee seniority points, and develops lists of certified candidates for various job openings within City gov- - emment. When a job opening occurs, the Board certifies candidates it deems qualified for the position to the City and the appropriate City official selects a candidate from the Board-certified list. See infra part I.D.
. Throughout this opinion, we refer to the pyramid-type hierarchy of the BFRS as follows: firefighter is the entry-level position within the department; fire lieutenant is the first significant promotion level to which a firefighter might aspire; fire captain is the next promotion following service as a fire lieutenant; battalion chief is the position to which a fire captain might be promoted; and fire chief is the top officer within the BFRS. We mean by the term firefighter only those BFRS employees occupying entry-level positions. Firefighter does not include fire lieutenants, captains, and battalion chiefs.
. Testimony indicates that the Mayor called the City Council to a special meeting the day before its regularly scheduled meeting to hear a presentation by attorneys for the City regarding a proposed settlement. It is unclear from the record whether Council members were given copies of the draft consent decree at that meeting. It is clear that two City attorneys involved in the litigation and settlement negotiations discussed the general terms of the decree with the Council and recommended settlement as the preferred course over continuing to defend the City’s record regarding employment opportunities for blacks and women.
. In 1981, when the decree was negotiated, approximately 9% of BFRS firefighters were black. This provision, therefore, meant that approximately 18% of annual promotions to lieutenant would be filled by blacks. The decree, however, also contained provisions designed to increase the number of black firefighters hired by the City. Therefore, as the percentage of black firefighters increased, the percentage of lieutenant promotions annually reserved for blacks would also increase.
.The record regarding back pay relief is somewhat confusing. At one stage in the negotiations, the United States was demanding that the City pay $500,000 in back pay relief to alleged victims of discrimination. The final decree provides for payment of $265,000 to settle all claims against the City. Whether the total amount of back pay relief to be provided by the City had been agreed upon when the decree was presented to the City Council is uncertain. Defense exhibit 39, the draft of the decree as it stood when the City Council met, contains a provision that "[t]he City agrees to pay the sum of $_ in full and complete settlement of the claims against the City of Birmingham for monetary relief of the blacks identified on Appendices_and_of this Decree." The Appellees maintain that attorneys for the City told the Ci1y Council that $265,000 would be paid to settle the claims.
James Baker, City Attorney when the decree was entered, testified at the 1991 trial. Baker's *1532testimony indicates that the Ci1y Council approved a settlement sum of $265,000 when it passed Resolution 547-81. Baker testified on cross examination as follows:
Q: And you prepared Resolution 547 which authorized entry into the .consent decree?
A: Yes, sir.
Q: And did Resolution 547 talk about the amount of money or just the entry into the decree?
A: My recollection is that — let me look at it and discuss the money because the City would have not been able to settle this case without council approving that sum of money being paid.
Q: Well, not only did the council approve the sum of money but the council approved entry into the decree by Resolution 547?
A: It did. Yes.
(Emphasis added.) Resolution 547-81, however, contains no reference to any sum of money to be appropriated for settlement. It simply authorizes the City Attorney, following consultation with the Mayor, to enter into a consent decree to settle pending litigation. We are unable to tell at what point the parties reached agreement on a dollar figure to settle outstanding claims. It is clear that no figure was approved in Resolution 547-81 and that the City desired to keep its financial obligations to a minimum.
.Certain aspects of the decree were modified by the district court in 1991. An appeal from.the modification order is before another panel of this Court. See Birmingham Firefighters v. Seibels, 11th Cir., 1994, 20 F.3d 1489.
. The decree also provides that, subject to the availability of quálified black candidates, at least one of the next two captain vacancies in the BFRS would be filled by a black, with subsequent captain and battalion chief openings filled at the rate of two times the percentage representation of blacks in the job category from which captains and battalion chiefs are usually drawn. ' Decree ¶ 8. It is unclear exactly how this provision was intended to work given that, at the time the decree was entered, there were no black lieutenants from which to select a black captain. Another provision of the decree, however, provides some insight: "Employees who have obtained permanent status as fire lieutenant or fire captain shall not be deemed ineligible for promotion to the next higher rank based upon any minimum length of service or time in rank.” Decree ¶ 19b. This allows black firefighters to progress rapidly through the ranks of lieutenant and captain to battalion chief.
. A district court need not await a party's motion to terminate a consent decree over which the court retains supervisory jurisdiction. When the remedy prescribed in the consent decree is accomplished, the district court may sua sponte *1533terminate the decree. United States v. City of Miami, 2 F.3d 1497, 1506 (11th Cir.1993). The district court has not yet acted to terminate the City decree.
. See also Local Number 93 v. City of Cleveland, 478 U.S. 501, 515, 106 S.Ct. 3063, 3072, 92 L.Ed.2d 405 (1986) (finding, for purposes of Title VII analysis, no need to distinguish between an employer's actions taken pursuant to a consent decree and voluntary action taken outside the litigation context); but see United States v. Paradise, 480 U.S. 149, 193-195, 107 S.Ct. 1053, 1078-79, 94 L.Ed.2d 203 (1987) (Stevens, J., concurring) (comparing the presumption against race-conscious decisionmaking a government must overcome when fashioning an affirmative action plan with the duty that a district court has to fashion a race-conscious remedy after finding a government guilty of racial discrimination and suggesting greater deference to the court ordered remedy).
. See also Mary C. Daly, Sbme Runs, Some Hits, Some Errors — Keeping Score in the Affirmative Action Ballpark from Weber to Johnson, 30 B.C.L.Rev. 1 (1988). Daly describes the opinions in the nine Supreme Court affirmative action cases preceding Croson as "lengthy, incohesive, contradictory, and ambiguous.... occupying] over five hundred fifty-four pages in the official reporters and consisting] of forty-six majority, plurality, concurring, and dissenting opinions." Id. at 5. Daly notes further that "[e]ven within a single case, it is often impossible to discern the Court’s holding because not every Justice in the majority will endorse the entire majority opinion” and frequently no majority opinion was produced at all. Id.
. This necessarily means that an employee with less seniority must have a higher converted score than a more senior employee in order to attain a higher final score and a higher ranking on the eligible register.
. The "rule of three” was in effect at all times pertinent to this appeal.
.This method was employed by the BFRS from the inception of the consent decree through 1988. In 1988, the BFRS began to use internal review panels to interview and evaluate the promotion candidates certified as eligible by the Board. This review panel system is not relevant to our inquiry because the City's actions about which Appellants here complain occurred before 1988 under the system described above.
. In Croson, a city attempted to remedy alleged discrimination in the construction industry by establishing minority business subcontractor set-asides for general contractors bidding on city construction projects. 488 U.S. at 476, 109 S.Ct. at 713. The city was not attempting to develop a remedy for its own past discrimination. Id. at 497, 109 S.Ct. at 724. This case, however, involves a city attempting to remedy under-representation of blacks and women that, the district court found, arose from the City’s own previous discriminatory employment practices, the evidence of which was before the City when it acted.
. See also United Steelworkers v. Weber, 443 U.S. 193, 204, 99 S.Ct. 2721, 2728, 61 L.Ed.2d 480 (1979) (observing, in a case involving a private employer, that "[i]t would be ironic indeed if a law triggered by a Nation’s concern over centuries of racial injustice and intended to improve the lot of those who had been excluded from the American dream for so long, constituted the first legislative prohibition of all voluntary, private, race-conscious efforts to abolish traditional patterns of racial segregation and hierarchy”) (citation omitted); id. at 200, 99 S.Ct. at 2726 (noting the " 'familiar rule that a thing may be within the letter of the statute and yet not within the statute, because not within its spirit nor within the intention of its makers' " (quoting Holy Trinity Church v. United States, 143 U.S. 457, 459, 12 S.Ct. 511, 512, 36 L.Ed. 226 (1892)).
. We address only the promotion provisions of the City plan in this appeal. Promotions, by their very nature, involve elevating a current employee to a higher position. Status as a department employee, therefore, is a prerequisite to promotion within a City department. Simply *1538put, an employee must first be a firefighter before becoming a fire lieutenant, a fire lieutenant before becoming a captain, and so forth.
Under the terms of the decree, the City cannot require a firefighter to serve longer than two years in that position before becoming eligible to take a promotional exam for fire lieutenant. Decree ¶ 19b. Further, the decree specifically removes all length of service and time-in-grade requirements for promotions to captain or above. Id. So, fire lieutenants promoted under the decree must have a minimum of two years' experience as a firefighter, but there is no minimum experience as a lieutenant required to become a captain, or as a captain in order to become a battalion chief.
. The Court noted that the agency director testified that Joyce’s sex was only one of numerous factors he considered in reaching the decision to promote her instead of Johnson. The Court likened the agency plan to the "Harvard Plan” approvingly discussed by Justice Powell in Regents of University of California v. Bakke, 438 U.S. 265, 315-318, 98 S.Ct. 2733, 2761-63, 57 L.Ed.2d 750 (1978) (approving consideration of an applicant’s race in a college admission program as a plus in the applicant's favor while not insulating the applicant from comparison with all other candidates for admission).
. The Equal Protection Clause provides, in relevant part, that "[n]o State shall make or enforce any law which shall ... deny to any person within its jurisdiction the equal protection of the laws." U.S. Const, amend. XIV, § 1.
. We discussed above and rejected Appellants’ assertion that, under Croson, the City was required to articulate specific findings of its past discrimination before entering the consent decree. See supra part III.A.1.
. We are unable to determine from the record how much of this progress to attribute to the effect of the City Council measure ordering department heads to set and achieve minority employment goals and how much to attribute to the curative measures the district court ordered the Board to take after finding that its entry-level tests had an adverse impact on blacks. We presume that the two worked in concert. ■