Billy Guice and Howard Claxton appeal the denial of their petition for writs of habeas corpus, which is based on their claim that blacks were excluded from service as foremen on the state grand jury that indicted them. This court, sitting en banc, remanded the case for an evidentiary hearing on the issue of racial discrimination in the selection of the grand jury foreman. Guice v. Fortenberry, 661 F.2d 496 (5th Cir.1981). That opinion sets out, and we shall not repeat, the facts and procedural history of the case up to that point. The court found that there was insufficient evidence in the record from which to determine whether the petitioners could state a prima facie case for discrimination under Castaneda v. Partida, 430 U.S. 482, 97 S.Ct. 1272, 51 L.Ed.2d 498 (1977). After the evidentiary hearing on remand, the district court found that the petitioners had made out a prima facie case, but that the State of Louisiana had successfully rebutted it. We reverse and remand with instructions that the writs of habeas corpus be issued ordering the indictments and convictions be set aside and providing the state the opportunity to seek new indictments and convictions.
I.
The purpose of the evidentiary hearing on remand was to allow the petitioners to produce evidence relating to the degree of underrepresentation of blacks as grand jury foremen, which was not presented at the state habeas hearing. This evidence is necessary under the three-part formulation of Castaneda under which the petitioners must:
(1) establish that the group against whom discrimination is asserted is a recognizable, distinct class, singled out for different treatment; (2) prove the degree of underrepresentation by comparing the proportion of the group in the total population to the proportion called to serve, here as foremen, over a significant period *278of time; and (3) support the presumption thus created by showing that the selection procedure is susceptible to abuse or is not racially neutral.
Guice v. Fortenberry, 661 F.2d at 499. The first and third of these elements had been established by the petitioners in their state court record. The degree of underrepresen-tation of blacks had not been established. Even though there was testimony that no black had been appointed grand jury foreman during the fifteen years prior to the time when the petitioners were indicted, there was no evidence of the total number of foremen appointed during those years. 661 F.2d at 504-06.
On remand, four witnesses testified at the evidentiary hearing. Three of them were called by the petitioners to establish their case of racial discrimination. The fourth witness was the state judge who selected the foreman of the grand jury that indicted the petitioners. The missing statistic was supplied at the hearing by the Chief Deputy Clerk of Court for Madison Parish: from the time that Judge Adams came on the bench in Madison Parish in October 1963, until the impanelling of the grand jury that indicted the petitioners, thirty-one foremen were chosen, twenty-eight of them by Judge Adams. The Chief Deputy Clerk testified that, to her personal knowledge, each of the appointed individuals was white. This comports with Judge Adams’ testimony at the state habeas hearing.
It was also established at the hearing that the grand jury venire of forty individuals, from which the judge selected the foremen, was randomly selected from a larger general venire of six hundred names. The primary source of names for the general venire was the voter registration roll of the parish.1 The jury commissions supplemented this list with the names of other qualified individuals in the community. A table showing the racial breakdowns of the voter registration rolls from 1967 through 1981 was entered into evidence. That table shows that forty-eight percent of the registered voters in Madison Parish in 1979 were black and that the percentage of black voters had been between forty-five and forty-nine since 1963.
The State of Louisiana presented Judge Cliff Adams as its sole witness. He had been a judge for the Sixth Judicial District, including Madison, East Carroll and Tensas Parishes, since October 1963. Under state law, it was his duty to select a grand jury foreman from the forty individuals who compose the grand jury venire. LA.CODE CRIM.PROC.ANN. art. 413 (West 1966). Judge Adams testified that he selected as foreman the individual whom he believed to be the best qualified for the position, regardless of race. His testimony revealed that the selection procedure was highly subjective, and that, for the most part, he relied on his personal knowledge of the qualifications of potential foremen. He admitted, “. .. if I don’t know them, I could easily leave out a qualified person... . ” Toward the end of his testimony, Judge Adams stated that, in retrospect, he felt he had chosen the most qualified foremen:
I was limited to my knowledge and I realize now that I’ve looked back over that list, yeah, there are Blacks in there that were qualified but I’ll say this, compared to the ones that I appointed, they weren’t better qualified. I will be willing to say the ones that I appointed were better qualified, White or Black.
Judge Adams testified that in the other two parishes in his jurisdiction he often made inquiries as to the qualifications of the members of the grand jury venire. Apparently he felt no need to do so in Madison Parish, where he resided and where he was often acquainted with members of the ve-nire. No evidence was presented of any systematic attempt to obtain objective information about the qualifications of venire members in Madison Parish.
*279II.
The magistrate found that the statistical evidence produced by the petitioners satisfied the second element of the Castaneda test, and that, coupled with evidence presented in the state habeas proceeding, it created a prima facie case of discrimination in the selection of the foreman of the grand jury that indicted the petitioners. The magistrate found, however, that the petitioners’ ease had been successfully rebutted by the state through the testimony of Judge Adams:
A prima facie case has been submitted by petitioners. As noted, it is determined from the demeanor of all the witnesses and of Judge Adams, that there certainly was no animosity at any time in appointing other than a non-black as foreperson of the grand jury. Certainly there was discrimination. However, there was no invidious discrimination and the judge merely was following the rules laid down by the Louisiana Supreme Court and selected the best foreperson he considered available of the grand jury.
The magistrate found that Judge Adams was justified in selecting foremen known personally to him, because “[h]e felt bound under Louisiana law ... to select the best person as grand jury foreman, who could stand up to the district attorney and if probable cause was not shown, to vote in that manner.”2 The magistrate noted Judge Adams’ pride in having appointed the first black jury commissioner and the first black grand jury foreman in the parish.
Over the objection of the petitioners, the district court adopted the findings and recommendations of the magistrate and denied the petitions for habeas relief. This appeal-was then taken.
III.
The issues on appeal in this case are whether the magistrate correctly determined that the petitioners had stated a prima facie case of discrimination and, if so, whether he was correct in holding that the State of Louisiana rebutted it.
A.
The state argues that the petitioners have failed to make out a prima facie case under Castaneda because underrepresentation has not been shown to have existed for a significant period of time. This, of course, is unfounded: the petitioners showed that no black had ever served as grand jury foreman in Madison Parish. There was personal testimony by a court official that each foreman from 1963 to 1979 was white. The state’s argument, however, is that'
the significant period of time to fairly judge the trial court’s choices of grand jury foreman as respects blacks and/or women under Castaneda, is no earlier than 1975 (when women were placed on the roles [sic]) and no earlier than 1977 (when a significant number of qualified blacks, i.e. teachers, ministers, officers, etc., were no longer excluded from the venires for known exemptions by law).
This argument proceeds from the premise that Judge Adams’ selection of foremen may be evaluated as an independent act only as to those years in which he had a racially nonbiased venire from which to choose a foreman. The state contends that the practice of automatically eliminating from the grand jury venire any individuals who would have been entitled to a statutory excuse from service had the effect of eliminating from the venire most blacks qualified for the leadership role of foreman.3 *280This practice, which the parish jury commissioners followed prior to 1976, was held unconstitutional in State v. Procell, 332 So.2d 814 (La.1976). There were only five opportunities for Judge Adams to select a foreman after the Procell decision, including the selection of the foreman of the grand jury that indicted the petitioners. Although no black foreman was selected on any of these five occasions, the state argues that the two-and-one-half-year period is simply an insufficient length of time to satisfy the Castaneda test.
The state urges us to hold that the petitioners have failed to state a prima fa-cie case simply because Louisiana has eliminated one admittedly discriminatory step in its foreman-selection process. We are not convinced that the distinction between the two periods (1963 to 1976 and 1976 to 1979) is so great that we are’ required to view them separately for the purpose of determining whether the petitioners have established a prima facie case. For one thing, the only evidence in the record tending to prove that the jury commission’s practice eliminated qualified blacks from the venire is the testimony of Judge Adams. He did not, however, say that the practice left the venires with no qualified blacks, nor did he actually testify that the percentage of qualified blacks in the venires had increased since 1976. Absent stronger evidence that the selection process has been materially changed, we have no trouble holding that the two periods of time should not be treated separately for the purpose of determining a prima facie case.
The petitioners have succeeded in proving a pattern of underrepresentation of blacks stretching back as far in time as any of the witnesses can remember. The record shows that over a period of fifteen years, during which the pool from which venire members were randomly selected was approximately forty-five percent black, no black grand jury foreman was appointed, even though there were thirty-one opportunities to do so. The fact of underrepresen-tation has clearly been established under Castaneda.
B.
The state is,- most certainly, entitled to rebut the presumption of discrimination established by the petitioners. To do so, it must show that the pattern of underrepre-sentation proved (no blacks appointed in fifteen years) was the result of a “racially neutral selection procedure.” Alexander v. Louisiana, 405 U.S. 625, 632, 92 S.Ct. 1221, 1226, 31 L.Ed.2d 536, 542 (1972).
The Supreme Court has held, on more than one occasion, that “affirmations of good faith in making individual selections are insufficient to dispel a prima facie case of systematic exclusion.” Id. See also Turner v. Fouche, 396 U.S. 346, 90 S.Ct. 532, 24 L.Ed.2d 567 (1970). In Hernandez v. State of Texas, 347 U.S. 475, 74 S.Ct. 667, 98 L.Ed. 866 (1954), the Supreme Court held that the testimony of five jury commissioners that they had tried to select the best qualified jurors was insufficient to overcome a prima facie case of denial of equal protection. The Hernandez court quoted with approval the following passage from an early equal protection case:
If, in the presence of such testimony as defendant adduced, the mere general as*281sertions by officials of their performance of duty were to be accepted as an adequate justification for the complete exclusion of negroes from jury service, the constitutional provision ... would be but a vain and illusory requirement.
Norris v. Alabama, 294 U.S. 587, 598, 55 S.Ct. 579, 584, 79 L.Ed. 1074, 1081 (1935).
The sweeping language of the relevant Supreme Court opinions does not foreclose the issue. A presumption of discriminatory conduct may be successfully rebutted by testimony of responsible public officials if that testimony establishes the use of racially neutral selection procedures. The Supreme Court said as much in Castaneda, where it noted the absence of testimony from grand jury commissioners as to the method by which they determined the qualifications for grand jurors. 430 U.S. at 499, 97 S.Ct. at 1282, 51 L.Ed.2d at 513. The question before us is whether the testimony of Judge Adams rises to the level required to rebut the petitioners’ case. We hold that it does not. Judge Adams’ testimony revealed that no objective criteria were used in his selection of grand jury foremen; rather, he selected individuals, always white, who were known to him. The dissent notwithstanding, Judge Adams’ testimony regarding the qualifications of the particular individual he chose as foreman of the grand jury does not undermine our reasoning when considered in the light of the fact that he testified that he made no inquiries regarding the qualifications of any of the other venire members. Judge Adams’ own evaluation of the nature of the selection process is revealed by his observation that “when we have a Black Judge, they’re gonna know who’s the most qualified amongst the Blacks to be a foreman.”
This case is distinguishable from United States v. Perez-Hernandez, 672 F.2d 1380 (11th Cir.1982), in which the selection of grand jury foremen in a Florida federal district court was upheld against an equal protection challenge.4 In Perez, a prima facie case of discrimination was established by the defendants, but was rebutted by the testimony of eight district judges involved in the foreman selection process.
Each judge testified that he acted independently of the other judges in choosing a grand jury foreman, although each employed similar guidelines in making a selection. These guidelines generally consisted of four separate factors: (1) occupation and work history; (2) leadership and management experiences; (3) length of time in the community; and (4) attentiveness during the jury empanelment. These factors directly relate to the ability to perform the administrative functions and duties of a grand jury foreman.
672 F.2d at 1387 (footnotes omitted).
The record in this case, unlike Perez, shows that there were no objective criteria or guidelines under which Judge Adams operated in his selection of grand jury foremen. Although his testimony was replete with “affirmations of good faith” in the performance of his duties, it also revealed that his approach to the selection process was necessarily subjective and biased. We believe, as do our colleagues on the Eleventh Circuit, that “testimony from ... alleged discriminators should be viewed with a great deal of judicial scrutiny.” United States v. Perez-Hernandez, 672 F.2d at 1387. Viewing the rebuttal evidence in this case, we hold it inadequate to overcome the presumption of discrimination established by the petitioners.
IV.
This decision should be the last from this circuit regarding the petition of Guice and Claxton, who have argued long and hard for the relief to which we today hold they are entitled. It has not been our task in this appeal and decision today to determine whether racial discrimination in the selection of a grand jury foreman requires that indictments be quashed and convictions be reversed. This result was the assumption *282of the Supreme Court in Rose v. Mitchell,5 and this court has since adopted the position that such discrimination compels voiding the indictments and convictions. “If convictions must be set aside because of taint of the grand jury, we see no reason to differentiate the result because discrimination affected the foreman.” Guice v. Fortenberry, 661 F.2d at 499. See Williams v. Mississippi, 608 F.2d 1021, 1022 (5th Cir.1979).6
Rather, it has been our task to determine whether in fact the petitioners were indicted by a grand jury whose foreman was chosen in a discriminatory manner. We hold simply that the petitioners, by showing a long history of the complete and absolute exclusion of blacks from a position for which approximately forty-five percent of the available candidates were black, have established a presumption of discrimination which has not been rebutted by the evidence adduced by the State of Louisiana.
We are advised by counsel that neither Guice nor Claxton is now in custody as a result of the constitutionally defective indictments. Nevertheless, the district court is instructed to issue writs of habeas corpus setting aside the indictments and the convictions of the petitioners, but providing, of course, that, the state may, within a reasonable period of time, seek to reindict the petitioners, and, if they are reindicted, seek to convict them.
REVERSED and REMANDED.
. See State v. Anderson, 315 So.2d 266 (La. 1975) (approving use of voter registration rolls as source of general venire).
. Article 413 of the Louisiana Code of Criminal Procedure provides that “the court shall select one person from the grand jury venire to serve as foreman of the grand jury.” The statute does not require or forbid any particular selection criteria.
. Former LA.CODE CRIM.PROC.ANN. art. 403 (West 1967) (amended 1975) provided:
The following persons are exempt from jury service, but the exemption is personal to them and is not a ground for challenge:
(1) The governor, lieutenant governor, state comptroller, state treasurer, secretary of state, superintendent of public education, their clerks and employees, the members, officers, and clerks of the legislature, and the judges and active officers of the several courts of this state;
*280(2) Any other public official, if jury service would seriously interfere with the performance of his official duties;
(3) Attorneys-at-law, peace officers, ministers of the gospel, physicians and dentists actively engaged in the practice of their professions, school teachers, school bus drivers, pharmacists, members of paid fire departments, chiefs and their first assistants of bona fide volunteer fire departments, and persons who are required to travel regularly and routinely in the course and scope of their employment;
(4) Persons who because of age, sickness, or other physical infirmity would suffer serious detriment if required to serve as a juror; and
(5) Persons who have served as grand or petit jurors in criminal cases or as trial jurors in civil cases during a period of twelve months immediately preceding their selection for jury service. '
Exemptions for prospective jurors are now governed by Louisiana Supreme Court rule.
. A part of the holding of Perez not relevant to this issue has been rejected by this circuit. United States v. Cronn, 717 F.2d 164, 167 (5th Cir.1983).
. 443 U.S. 545, 553 n. 4, 99 S.Ct. 2993, 2998 n. 4, 61 L.Ed.2d 739, 747 n. 4 (1979).
. In this case, because the foreman was selected from the venire rather than from the grand jury itself, any discrimination in the selection of a foreman necessarily tainted the composition of the grand jury as well: only eleven of its twelve members were picked at random.