HARRISON L. WINTER, Chief Judge:

Both the Equal Employment Opportunity Commission (EEOC) and Liggett & Myers, Inc. (the company) appeal from the judgment entered by the district court in a suit instituted by the government 1 against the company under the Equal Pay Act provisions of the Fair Labor Standards Act, as amended, (FLSA) 29 U.S.C. §§ 201-19. The district court found that the company had violated § 206(d) of the FLSA by paying twenty-two female employees in its Durham, North Carolina plant less than it paid male employees doing comparable work. Back pay with prejudgment interest was awarded to each of the injured employees.

The company challenges the finding that it violated the FLSA in relation to three of the twenty-two female employees, and also claims that the district court erred in setting the rate of prejudgment interest. In addition, EEOC and the company each challenge aspects of the method used by the district court to calculate the amounts of the back pay awards. We reject the company’s challenges to the finding that it violated the FLSA and to the setting of the rate of prejudgment interest, but we conclude that the district court erred in calculating the amounts of the back pay awards. We affirm the adjudication of liability but vacate the awards and remand the case for reeomputation of back pay.

We consider first the company’s challenge to its liability to Mary Hilliard, Ruby Olive and Ruth Fowler.

Hilliard was employed as a production supervisor in the Durham plant’s manufacturing department during the relevant time period. Like every other female production *1074 supervisor in the department, she was paid less in every relevant year than the lowest paid male production supervisor in the department, yet the district court found that every production supervisor in the department performed comparable work.

Olive was one of three production supervisors employed in the Durham plant’s blending department during the relevant time period. She and a male production supervisor named W. E. Lucas were promoted to their positions at roughly the same time as a male production supervisor named H. P. Carpenter retired. Carpenter was paid considerably more than Olive or Lucas, who were paid the same amount. The district court found that Olive and Carpenter performed comparable work, but that Olive and Lucas did not.

Fowler was one of two production supervisors employed in the Durham plant’s returned goods department during the relevant time period. She was promoted to the position at roughly the same time as a male production supervisor named James Phelps retired. Phelps was paid considerably more than Fowler, yet the district court found that Fowler and Phelps performed comparable work.

To summarize, the district court found that Hilliard, Olive and Fowler were paid less than male employees doing comparable work. The company does not dispute these findings, and it acknowledges that the burden was on it to show that the disparate treatment resulted from consideration by the company of legitimate factors. Despite the district court’s contrary finding, the company argues that it met the burden, pointing to testimony by one of its managers that hourly employees who were promoted to supervisory positions — including Hilliard, Olive and Fowler — were always given a yearly salary equal to the yearly value of their hourly wages plus a supplement of several hundred dollars. This evidence does not prove disparate treatment from legitimate factors. It was not shown that the hourly wages these female employees received before their promotions were not sexually discriminatory and that it was not harder for women than for men to become supervisors without first being hourly employees. Thus we cannot say that the finding of liability to Hilliard, Olive and Fowler was clearly erroneous.

II.

It was entirely proper for the district court in the exercise of sound discretion to award prejudgment interest to make the injured female employees whole. See Marshall v. Board of Education, 470 F.Supp. 517 (D.Md.1979), aff’d without opinion, 618 F.2d 101 (4 Cir. 1980). In determining the rate of prejudgment interest, the district court is not bound by state law. See, e.g., General Facilities, Inc. v. National Marine Service, Inc., 664 F.2d 672, 674 (8 Cir. 1981). That does not mean, however, that the district court may not in its discretion choose to apply the interest rate provided for by state law. In the present case, EEOC asked for prejudgment interest at the rate of eight percent, the legal rate currently provided for in § 24-1 of the North Carolina Code, and the company responded by citing North Carolina cases in which the court had awarded prejudgment interest at the rate of six percent, the legal rate provided for by § 24-1 prior to its most recent amendment. The district court apparently decided to resolve the dispute by referring to state law. It pointed out to the company that the cases it had cited had been decided prior to the amendment of § 24-1, and agreed with EEOC that prejudgment interest should be assessed at the rate of eight percent.

On appeal, the company argues that the portion of the interest accruing before July 1, 1980, the effective date of the amendment, should have been calculated at the rate of six percent. In light of the election apparently made by the district court, we think that the matter should now be decided by reference to state law, and we conclude that the district court applied North Carolina’s law correctly. The bill which amended § 24-1 to change the legal rate from six to eight percent provided that "this act shall not apply to judgments en *1075 tered prior to July 1, 1980.” 1979 N.C.Sess. Laws, 2d Sess., chap. 1157, § 8 (emphasis added). The clear implication of this language is that the new rate of interest shall apply to all of the interest awarded in judgments entered on July 1, 1980, or thereafter, instead of just to that portion of the interest accruing after the amendment becomes effective. Accordingly, we reject the company’s challenge to the setting of the rate of prejudgment interest.

III.

The method used by the district court to calculate the amounts of the back pay awards was set forth in the district court’s opinion filed in December 1979 and in an order filed in November 1980. In the former, the district court stated:

The Secretary [of Labor] also suggested at several points in his proposed findings of fact and conclusions of law that the female employees found by the Court to be owed additional wages should be reimbursed only for the difference between their actual wages and the average salary paid to comparable male workers. The Secretary cited no authority for this contention. There appears to the Court no reason to limit the compensation paid to female workers to the average male salary. A logical extension of the Secretary’s position would be that an employer could achieve future compliance with the Act by paying all workers at the average male salary. Unless all males had been paid the same salary in the past, this course of action would directly contravene the prohibition in [§ 206(d)] that bars employers from complying with the Act by reducing the wage of any employee. See 29 U.S.C. § 206(d)(1). A more logical outcome of the Court’s finding that there was no legal justification for the pay differentials between male and female employees would seem to be that female workers are entitled under the Act to be paid the salary received by the highest paid comparable male worker for whom the defendant has failed to establish a legitimate reason for his wage differential.

In the November 1980 order the district court stated:

NOW THEREFORE, IT IS ORDERED, ADJUDGED AND DECREED AS FOLLOWS:

1. Withheld wages due females in comparable jobs to males shall be calculated up to the time the female salary equals the salary of the comparable male or up until such time as she quit, retired, was terminated or transferred to a non-comparable job, whichever occurred first.

2. If the comparable male quit, retired, was terminated or transferred to a non-comparable job, the female’s withheld wages shall be calculated up to the time her salary equals the salary of the comparable male at the time he quit, retired, was terminated or transferred.

The company argues that, instead of comparing each female employee’s salary with that of the highest paid male employee doing comparable work, the district court should have made individualized back pay awards based on each female employee’s seniority and education or, where that was not feasible, should have compared each female employee’s salary with the average salary paid to male employees doing comparable work. EEOC, on the other hand, argues that in instances where the male employee used for comparison left the company or was transferred prior to the end of the relevant time period, while the corresponding female employee continued to do the same job, the district court erred by basing the back pay award for the remainder of the relevant time period on the salary paid to the male employee at the time he ceased doing comparable work. EEOC points out that it was customary for the company’s male employees to receive periodic raises, and argues that the district court should have made an allowance for any raises the male employee would have received if he had continued to do the comparable work. 2 We agree with the essence of both sides’ criticisms.

*1076 A.

A district court admittedly has wide discretion in the formulation of a remedy in a case of this sort, and EEOC argues that it was a proper exercise of discretion for the trial judge both to decline to attempt individualized back pay awards and to decline to base the back pay awards on a comparison with the average salary paid to male employees doing comparable work. EEOC submits that the trial judge declined to do the former after considering the matter and determining that such an approach would be unworkable, and that the district judge declined to do the latter after considering the matter and determining that such an approach would result in an unacceptable risk of under-compensation. We disagree.

With respect to individualized back pay awards, there is no indication that the district court even considered their possibility. EEOC argues that the possibility was ruled out by the district court’s findings that consideration by the company of legitimate factors could not explain the spread between female salaries and male salaries. For individualized back pay awards to be feasible, however, it is only necessary that consideration by the company of legitimate factors tends to explain the spread among various male salaries. Proof that female workers were paid less than male workers doing comparable work, even though the female workers had more education and seniority, simply does not establish that individualized back pay awards are not feasible. It is true that EEOC has cited examples where even the spread among various male salaries does not seem explainable on the basis of education or seniority, but the district court also gave examples where it does. More importantly, counsel for EEOC conceded at oral argument that for many of the twenty-two injured female employees, a male employee doing comparable work and possessing comparable seniority and education could be identified. Where that can be done and where as here we are dealing with only twenty-two injured employees, we conclude that the female employee should be awarded the difference between what she was paid and what the comparable male employee was paid.

We, of course, do not rule that individualized back pay awards are required in all cases or for all employees. In many situations, a district court may properly conclude that the approach would be unworkable — to name a few, where the number of affected employees is so large as to render individualized back pay awards unduly burdensome on the judicial process, where there is an absence of comparables, etc. Given the facts in the present case, however, we think that the case must be remanded so that the district court may fully consider the possibility of applying an individualized approach. While the approach may prove unworkable for the injured female employees in the smaller departments, we are inclined to think that the approach will prove workable for the bulk of the injured female employees.

In those instances, if any, where the individualized approach is shown not to be feasible, it will be necessary for the district court to choose between using the highest male salary or the average male salary as the basis for comparison. We see no need to pass on EEOC’s contention that, in certain fact situations, a district court could properly conclude that using the average male salary would result in an unacceptable risk of under-compensation. 3 We perceive *1077 no basis for such a conclusion in the present case, and in any event it is clear that the district court did not, as EEOC contends, arrive at such a conclusion. Rather, the district court declined to use the average male salary on the basis of an obviously fallacious argument that if paying female employees the average male salary is a proper remedy, then it would be proper to pay all employees the average male salary, but paying all employees the average male salary would be illegal since some males would have their salaries cut and the FLSA prohibits cutting anyone’s salary as part of a remedy. The fallacy in this argument, of course, is that the second proposition simply does not follow from the first. We conclude that in those instances where the individualized approach does not prove workable, each female employee’s wages should be compared with the average salary paid to male employees doing comparable work. 4

Even if the district court is able on remand to identify a comparable male employee for each of the female employees entitled to an award of back pay, there may be instances where that male employee left the company or was transferred prior to the end of the relevant time period. The district court ruled that, in those instances, the salary paid to the male employee at the time he ceased doing comparable work should continue to be the benchmark against which the female employee’s salary would be compared for the remainder of the relevant time period. 5 In effect, this amounted to an assumption that, if the male employee had continued doing comparable work until the end of the relevant time period, he would have received no salary increases. In this case, the evidence clearly established that the company routinely gave yearly salary increases to male *1078 employees — a fact which the district court itself acknowledged. The district court’s implicit assumption thus was unrealistic. We conclude that if the male employee used for comparison ceased doing comparable work prior to the end of the relevant period, the district court should determine the salary increases which that employee would have received had he continued to do comparable work until the end of the relevant time period. The corresponding female employee should then be compensated for these salary increases. 6

While we are aware of no court of appeals decision adopting this approach in an FLSA case, 7 we note that we have adopted a similar approach in Title VII cases, see EEOC v. Korn Industries, Inc., 662 F.2d 256, 263 (4 Cir. 1981); White v. Carolina Paperboard Corp., 564 F.2d 1073, 1084 (4 Cir. 1977); United Transportation Local 974 v. Norfolk & Western Ry. Co., 532 F.2d 336, 341 (4 Cir. 1975), cert. denied, 425 U.S. 934, 96 S.Ct. 1664, 48 L.Ed.2d 175 (1976). Given that the back pay awards were reduced as a result of any salary increases the female employees received after males ceased doing comparable work, the female employees can only be made whole if they are compensated for any salary increases the males would have received if they had not left the company or been transferred.

AFFIRMED IN PART; VACATED IN PART; AND REMANDED.

. The suit against the company was filed in 1974 by the Department of Labor, which prosecuted the suit to final judgment in the district court in November 1981. In December 1981, EEOC was substituted as plaintiff pursuant to a government reorganization. EEOC briefed and argued the cross-appeals in this court.

. Given the other aspects of the district court’s method for calculating the amounts of the back *1076 pay awards, the error alleged by EEOC reduced the amount of the back pay awards received by ten of the twenty-two injured employees.

. EEOC cites three cases which it says indicate that using the highest male salary is proper. See Syvock v. Milwaukee Boiler Mfg. Co., 665 F.2d 149, 160-61 (7 Cir. 1981); EEOC v. Detroit Edison Co., 515 F.2d 301, 315 (6 Cir. 1975), vacated and remanded on other grounds, 431 U.S. 951, 97 S.Ct. 2669, 53 L.Ed.2d 267 (1977); Bowe v. Colgate-Palmolive Co., 489 F.2d 896, 902 (7 Cir. 1973).

Syvock was decided under the Age Discrimination in Employment Act (ADEA), 29 U.S.C. § 621-34, but monetary awards in ADEA cases are generally governed by the law developed under the FLSA, see Spagnuolo v. Whirlpool Corp., 641 F.2d 1109, 1114 (4 Cir.), cert. denied, 454 U.S. 860, 102 S.Ct. 316, 70 L.Ed.2d 158 *1077 (1981); Slatin v. Stanford Research Institute, 590 F.2d 1292 (4 Cir. 1979). In calculating a back pay award, the district court in Syvock assumed that, absent age discrimination, plaintiff would have earned the highest hourly rate paid to any employee doing comparable work. In calculating the award, however, the district court used the plaintiffs own absenteeism record and the average overtime record of all the company’s employees. Accordingly, the award was considerably less than the difference between the plaintiff’s actual wages and the wages of the highest paid worker doing comparable work. The company did not appeal, so the court of appeals was not faced with the question of whether the comparison to the highest hourly rate was proper. Plaintiff appealed, claiming that he should have received the full difference between his actual wages and the wages of the highest paid worker doing comparable work. The court of appeals rejected this argument, stating: “It would be wholly inappropriate to reward the plaintiff, and penalize the employer, for the extraordinary diligence and motivation of other workers.” 665 F.2d at 161. The case is hardly compelling authority for the EEOC’s position.

In Detroit Edison, a Title VII race discrimination case, the court said that each of the injured workers was entitled to “the difference between [his actual earnings] and the highest earnings for the jobs he would have bid on and qualified for if a non-discriminatory system of seniority, job posting and bidding and transfer had been in effect.” 515 F.2d at 315 (emphasis added). In Bowe, a Title VII sex discrimination case which the Detroit Edison court cited to support the holding just quoted, the court said that each injured worker was entitled to the difference between her actual rate of pay and “ ‘the highest rate of pay for such jobs as [the employee] would have bid on and qualified for if a non-discriminatory seniority scheme would have been in existence.’ ” 489 F.2d at 902 (emphasis added) (quoting Bowe v. Colgate-Palmolive Co., 416 F.2d 711, 721 (7 Cir. 1969) (before remand)). If it is assumed that the pay for any one “job” was uniform — which appears likely — Detroit Edison and Bowe stand for the exact opposite of the proposition for which EEOC cites them, because they limit each plaintiff to the jobs he or she could have qualified for.

. It is worth noting that this is the position that even the government took in the present case until the district court, acting on its own initiative, concluded to use the highest male salary as the basis of comparison.

. As the district court viewed the matter, the male employee to be used for comparison was the highest paid male employee doing comparable work. In our view, on the other hand, the male employee to be used for comparison would be either (1) a male employee doing comparable work and possessing comparable education and seniority or (2), in the event an individual described in (1) could not be identified, a hypothetical male employee earning the average salary paid to male employees doing comparable work. See Section III A supra.

. By regulation, the Secretary of Labor (whose enforcement duties have since been transferred to EEOC) has indicated that an employer violates § 206(d) by using men to perform a given job, replacing them with women, and paying the women less to do the same job, even if the employer never uses men and women simultaneously to perform comparable work. See 29 C.F.R. § 800.114(c). By negative inference, this seems to indicate that no violation would occur if the employer used men to perform a given job, replaced them with women, and paid the women exactly as much as the men had been making at the time the job was turned over to the female employees. Indeed, the regulation itself states: “Compliance with the Act . . . can be achieved only by increasing the wage rate [paid to the female employees] to the higher rate for the job when performed by employees of the opposite sex.” Id. By analogy, it can be said that the aspect of the district court’s calculation method which we reject in this section of our opinion was endorsed by the regulation. However, the regulation was promulgated and amended for the only time in early 1967, see 32 Fed.Reg. 2380 (1967); 32 Fed.Reg. 5737 (1967), before the onset of pronounced and prolonged inflation. We do not think that the regulation precludes the government, in a case where the evidence establishes that the employer made a practice of granting yearly salary increases to male employees, from advocating the method of calculating back pay awards which we adopt here.

. The decisions we are aware of in which a court of appeals has enforced 29 C.F.R. § 800.-114(c) are all inapposite. Either the defendant prevailed below, so that there was no calculation of a back pay award, and the court of appeals remanded without any specific instructions on how a back pay award should be calculated, or neither party challenged the exact amount of a back pay award entered below. See Marshall v. School Board, 599 F.2d 1220, 1222-23 (3 Cir. 1979); Pearce v. Wichita County, 590 F.2d 128, 133 (5 Cir. 1979); Peltier v. City of Fargo, 533 F.2d 374, 377 (8 Cir. 1976); Hodgson v. Behrens Drug Co., 475 F.2d 1041, 1049 (5 Cir.), cert. denied, 414 U.S. 822, 94 S.Ct. 121, 38 L.Ed.2d 55 (1973); Hodgson v. Miller Brewing Co., 457 F.2d 221, 226-27 (7 Cir. 1972).