After uncovering a large and violent crack cocaine conspiracy in the Williamsburg, Virginia area, the United States indicted 25 persons in three separate indictments. All 25 are black, and of the approximately 55 unindicted persons who were also involved in the conspiracy, 50 are black and 5 are white.

Two of the blacks who were indicted, Anthony L. Olvis and Angela D. Palmer, moved to dismiss their indictment, claiming that they had been selected for prosecution because of their race in violation of the equal protection component of the Fifth Amendment's Due Process Clause. On Olvis' and Palmer's argument that five unindicted white persons were similarly situated to them and their demonstration that more than 90% of those indicted in the Norfolk-Newport News area since 1992 for crack cocaine trafficking are black, the district court ordered discovery into the government's criteria for prosecution. When the government refused, arguing that such discovery was unwarranted, the district court dismissed the indictment against Olvis and Palmer.

Because the defendants did not satisfy the "rigorous" standard recently articulated by the Supreme Court to obtain discovery for selective-prosecution claims, we conclude that the district court erred in ordering discovery. Accordingly, we reverse the court's dismissal order and remand with instructions to reinstate the indictment and for further proceedings.

* In 1992, drug enforcement officers began investigating Anthony Olvis and other individuals involved in distributing crack cocaine and marijuana in Williamsburg, portions of James City and York Counties, and an area of Newport News, Virginia. Investigators believed that Olvis had managed to become "untouchable," insulating himself from prosecution by employing a network of street dealers whom he replaced as they were arrested. Moreover, according to law enforcement authorities, Olvis' activities had grown increasingly violent, with shootings becoming "mixed in with the street level activity." Indeed, at the time of the indictment in this case, police believed that Olvis' drug conspiracy was linked to several unsolved murders. The government accordingly claims that it had a keen interest in prosecuting Olvis and his colleagues.

Of more than 80 persons allegedly involved in the drug trafficking activities, the government prosecuted 25 through 3 indictments.* Olvis and Angela Palmer, who assisted Olvis in laundering his drug proceeds by taking title to vehicles in her name, were indicted in August 1995 for conspiracy to distribute crack cocaine and money laundering. Olvis was also indicted for possession with intent to distribute crack cocaine and use of a firearm in relation to drug trafficking. Palmer was indicted additionally for perjury and obstruction of justice.

In support of their motion to dismiss on selective-prosecution grounds, Olvis and Palmer noted that all 25 defendants that the government indicted are black. While recognizing that "a number of people that have been granted immunity are also black," the defendants argued that grand jury testimony revealed that "at least three whites have been granted immunity who are at least or more culpable than the black defendants, and at least two other whites have not been charged."

In opposition to the dismissal motion, the government insisted that the unindicted white individuals, "like the many unindicted black persons Olvis neglect[ed] to list," were not similarly situated to Olvis or Palmer. The government's chief prosecutors submitted two affidavits, stating that the government's decisions about whom to indict were based "solely on the evidence or other race-neutral criteria such as a prior disposition in state court, more limited participation in the conspiracy, or the government's need for a particular individual's cooperation."

At the hearing on the motion, counsel for Olvis and Palmer recited statistical evidence that more than 90% of those who had been tried since 1992 for crack cocaine offenses in the district court's Norfolk and Newport News divisions are black. They claimed that the defendant's race was known in 285 of the 312 crack cocaine cases tried since 1992, and that 260 of those 285 cases involved black defendants.

The government presented testimony from Kevin Comstock, an Assistant United States Attorney, and Lieutenant Delmas Linhart, an officer in the James City County Police Department and special deputy with the FBI. In his testimony Comstock indicated that the percentage of blacks indicted for crack cocaine offenses was high because blacks primarily were involved in the distribution of crack cocaine in the Norfolk-Newport News area. He explained,

I don't choose individuals to violate the law. They choose to violate the law themselves. And when they violate the law, if we can prove it, we prosecute[ ] them, regardless of their race, regardless of their sex, regardless of where they were born, or in what family they were raised in.

Lt. Linhart testified similarly that race played no part in the government's selection of whom to prosecute. He noted that the Colonial Narcotics Enforcement Task Force, of which he was a member, pursued the Olvis organization because Olvis had managed to insulate himself from the police and his organization was becoming increasingly violent. And Lt. Linhart added that Angela Palmer had been offered the opportunity to become a cooperating witness, but that, in the investigators' judgment, she had lied before the grand jury and had otherwise failed to cooperate.

Lt. Linhart also explained why each white conspirator cited in the defendants' motion had not been indicted. Mary Deroja, who was romantically involved with an alleged gang member and drove his car for him, had approached the authorities and agreed to assist in their investigation by working undercover. Denny Petrie had not yet been prosecuted because the government, after executing a search warrant on his home, still lacked sufficient evidence to indict him. Lonnie Beverly and his black partner, Eddie Phillips, had been approached about becoming witnesses at a time when the authorities knew little about them; only subsequently did enforcement officers learn that Beverly was the driver and Phillips the gunman in a drive-by shooting of an unoccupied car, information that might have disqualified them from receiving immunity. Lt. Linhart testified, moreover, that Beverly has since proven truthful and cooperative. As for the remaining two unindicted white persons, Linhart testified that the government had never heard of "Floyd," a white male referred to once in an ambiguous manner by a witness before the grand jury, nor did it have any evidence that Jeffrey Branscome was a seller, rather than merely a user, of drugs.

The district court concluded that the defendants had made a "nonfrivolous showing in raising a claim of selective prosecution," and ordered the government to respond to defendants' formal requests for discovery into its criteria for selecting whom to prosecute. The court found that Olvis, Palmer, and the unindicted white conspirators were "similarly situated" for a selective-prosecution claim and that the defendants' statistical data made a "nonfrivolous showing of discriminatory intent."

The government moved for reconsideration of the district court's order, submitting an additional affidavit from Lt. Linhart which detailed the shootings and murders associated with the "largest drug ring in Hampton Roads' history." Lt. Linhart's affidavit pointed out that all but one of the shootings' victims were black and that there were more than ten unindicted black conspirators for every unindicted white conspirator. The affidavit added that after the indictments had been publicly announced, Lt. Linhart received several phone calls from black citizens in the predominantly black neighborhoods in which the gangs operated, expressing their appreciation for the arrests.

The district court denied the government's motion for reconsideration. And when the government refused to comply with its discovery order, the court dismissed the indictment against Olvis and Palmer.

This appeal followed.

When acting on probable cause that a crime has been committed, a government prosecutor generally enjoys unfettered discretion in the decision whether to prosecute. See Bordenkircher v. Hayes, 434 U.S. 357, 364, 98 S. Ct. 663, 668-69, 54 L. Ed. 2d 604 (1978). Because law enforcement is "a core executive constitutional function," United States v. Armstrong, --- U.S. ----, ----, 116 S. Ct. 1480, 1486, 134 L. Ed. 2d 687 (1996), judicial intrusion into prosecutorial decisions is justified only when the Constitution requires it, see United States v. Batchelder, 442 U.S. 114, 125, 99 S. Ct. 2198, 2205, 60 L. Ed. 2d 755 (1979).

The equal protection component of the Fifth Amendment's Due Process Clause forbids the government from deciding to prosecute based on a defendant's race. See Armstrong, --- U.S. at ----, 116 S. Ct. at 1486; see also Oyler v. Boles, 368 U.S. 448, 456, 82 S. Ct. 501, 505-06, 7 L. Ed. 2d 446 (1962). Because of the great danger of unnecessarily impairing the performance of a core executive constitutional function and the "presumption of regularity" that supports prosecutorial decisions, a defendant must support a selective-prosecution claim with "clear evidence." Armstrong, --- U.S. at ----, 116 S. Ct. at 1486 (quoting United States v. Chemical Found., Inc., 272 U.S. 1, 14-15, 47 S. Ct. 1, 6, 71 L. Ed. 131 (1926)). The standard is intended to be a "demanding" and "rigorous" one. Id. at ----, ----, 116 S. Ct. at 1486, 1488.

To establish a selective-prosecution claim, a defendant must demonstrate that the prosecution "had a discriminatory effect and that it was motivated by a discriminatory purpose." Wayte v. United States, 470 U.S. 598, 608, 105 S. Ct. 1524, 1531, 84 L. Ed. 2d 547 (1985). This requires the defendant to establish both (1) that "similarly situated individuals of a different race were not prosecuted," Armstrong, --- U.S. at ----, 116 S. Ct. at 1487, and (2) that the decision to prosecute was "invidious or in bad faith," United States v. Greenwood, 796 F.2d 49, 52 (4th Cir.1986) (quoting United States v. Berrios, 501 F.2d 1207, 1211 (2d Cir.1974)).

Similarly, to obtain discovery in support of a selective-prosecution claim, a defendant must produce "some evidence" making a "credible showing" of both discriminatory effect and discriminatory intent. Armstrong, --- U.S. at ---- - ----, 116 S. Ct. at 1488-89. Just as the standard for ultimately proving a selective prosecution claim is a rigorous one, so too is the evidentiary threshold for obtaining discovery from the government to support such a claim. Id. at ----, 116 S.Ct. at 1488. A significant barrier to discovery is necessary because discovery "imposes many of the costs present when the Government must respond to a prima facie case of selective prosecution"; it diverts governmental resources and discloses prosecutorial strategies. Id. Thus, the Supreme Court recently admonished in Armstrong, "the showing necessary to obtain discovery should itself be a significant barrier to the litigation of insubstantial claims." Id. at ----, 116 S. Ct. at 1486.

While we ordinarily defer to a district court's decision to order discovery, the Supreme Court emphasized in Armstrong that for a selective prosecution claim, the justifications for a rigorous standard of proof to establish a claim require "a correspondingly rigorous standard for discovery in aid of such a claim." --- U.S. at ----, 116 S. Ct. at 1488. Thus, when we review a district court's discovery order in support of a selective-prosecution claim, we are determining the legal adequacy of the evidence. We review the legal adequacy of evidence de novo. Cf. Ornelas v. United States, --- U.S. ----, ----, 116 S. Ct. 1657, 1663, 134 L. Ed. 2d 911 (1996) (legal adequacy of evidence to establish probable cause is reviewed de novo ); Towler v. Sayles, 76 F.3d 579, 581 (4th Cir.1996) (legal adequacy of evidence on appeal from judgment as a matter of law is reviewed de novo ).III

The district court in this case concluded that Olvis and Palmer had satisfied the discriminatory effect prong of their selective-prosecution claim because the unindicted white conspirators were similarly situated to them. Focusing only on the evidence of the conspirators' criminal activity, the court found all to have "similar involvement" in the conspiracy and dismissed the government's proffered distinctions among the conspirators as "unpersuasive." Thus, the court reasoned that the government's inconsistency in granting some members of the conspiracy immunity and prosecuting others "call[ed] into question the legitimacy of the entire prosecution."

Generally, in determining whether persons are similarly situated for equal protection purposes, a court must examine all relevant factors. See Ah Sin v. Wittman, 198 U.S. 500, 507-08, 25 S. Ct. 756, 758-59, 49 L. Ed. 1142 (1905); see also Attorney General of the United States v. Irish People, Inc., 684 F.2d 928, 946 (D.C.Cir.1982) ("Discrimination cannot exist in a vacuum; it can be found only in the unequal treatment of people in similar circumstances"), cert. denied, 459 U.S. 1172, 103 S. Ct. 817, 74 L. Ed. 2d 1015 (1983). And this principle applies to the analysis of allegedly discriminatory criminal prosecutions:

The goal of identifying a similarly situated class of law breakers is to isolate the factor allegedly subject to impermissible discrimination.... If all other things are equal, the prosecution of only those persons [to whom the factor applies] ... gives rise to an inference of discrimination. But where the comparison group has less in common with defendant, then [other] factors ... may very well play a part in the prosecution.

United States v. Aguilar, 883 F.2d 662, 706 (9th Cir.1989).

By considering only the conspirators' relative culpability, the district court in this case failed to take into account several factors that play important and legitimate roles in prosecutorial decisions. See Oyler, 368 U.S. at 456, 82 S.Ct. at 505-06 (prosecutorial exercise of selectivity in enforcement without more does not violate Constitution). A prosecutor may, for example, legitimately offer immunity to a potential defendant who is equally "involved" as others simply because that defendant may choose to cooperate and expose yet more criminal activity. Prosecutorial decisions may also be legitimately influenced by such factors as the strength of the evidence against a particular defendant, the defendant's role in the crime, whether the defendant is being prosecuted by state authorities, the defendant's candor and willingness to plead guilty, the amount of resources required to convict a defendant, the extent of prosecutorial resources, the potential impact of a prosecution on related investigations and prosecutions, and prosecutorial priorities for addressing specific types of illegal conduct.

Making decisions based on the myriad of potentially relevant factors and their permutations require the very professional judgment that is conferred upon and expected from prosecutors in discharging their responsibilities. See McCleskey v. Kemp, 481 U.S. 279, 297, 107 S. Ct. 1756, 1769-70, 95 L. Ed. 2d 262 (1987). In rejecting the district court's narrow approach to relevant factors to be considered when deciding whether persons are similarly situated for prosecutorial decisions, we hold that defendants are similarly situated when their circumstances present no distinguishable legitimate prosecutorial factors that might justify making different prosecutorial decisions with respect to them.

Applying our fact-focused test to the circumstances of this case, we conclude that the defendants did not establish the first element of their selective-prosecution claim. We begin with the observation that Olvis is alleged to be the head of an organization that has distributed cocaine continuously for over four years and has become increasingly violent. While Palmer does not project the same prosecutorial profile, she is being prosecuted because investigators in this case believe that she committed perjury before the grand jury and she has otherwise refused to cooperate with law enforcement authorities.

Compared to Olvis and Palmer, the white conspirators projected significantly different prosecutorial profiles. Mary Deroja came forward and offered to go undercover to assist the investigation, and in doing so she offered to put herself at significant risk. Although Denny Petrie was a government target, a search of his residence failed to produce sufficient evidence to prosecute him. Lonnie Beverly, who is white, along with Eddie Phillips, his black partner, agreed to become witnesses before the government uncovered the extent of their illegal activities. Nevertheless, Beverly continues to be truthful and helpful to the prosecution. And the government has no information about a white person named Floyd, whose last name is unknown, nor evidence that Jeffrey Branscome was a drug dealer. Moreover, in addition to presenting these particularized factors that distinguish the defendants' circumstances from those of the unindicted conspirators, the government notes that there were approximately 50 black conspirators whom prosecutors chose not to indict. We find no merit, based on all of these legitimate considerations, to the contention that Olvis and Palmer were situated similarly to the unindicted white conspirators for making prosecutorial decisions.

As further evidence of their argument that the government's prosecutorial decisions had a discriminatory effect, Olvis and Palmer presented an informal statistical study conducted by their attorneys. The study showed that in the Norfolk-Newport News area, of all the federal crack cocaine trafficking prosecutions in federal court since 1992 in which the defendant's race was apparent, over 90% involved black defendants.

While the district court viewed the defendants' informal study as probative of only the discriminatory intent prong of a selective-prosecution claim, the Supreme Court in Armstrong considered a similar study under the discriminatory effect prong. See --- U.S. at ---- - ----, 116 S. Ct. at 1488-89. Regardless of the purpose for which the defendant's study is offered, however, we do not believe that it advances the defendants' selective-prosecution claim.

Olvis' and Palmer's study provides no statistical evidence on the number of blacks who were actually committing crack cocaine offenses or whether a greater percentage of whites could have been prosecuted for such crimes. The study suffers from the same fatal defect that the Court recognized in Armstrong, where the proffered study had failed "to identify individuals who were not black, could have been prosecuted for the offenses for which respondents were charged, but were not so prosecuted." --- U.S. at ----, 116 S. Ct. at 1489. Without an appropriate basis for comparison, raw data about the percentage of black crack cocaine defendants proves nothing. Such statistics could have relevance only if it could be presumed that crack cocaine violations were committed proportionately by all races--a presumption the Supreme Court rejected in Armstrong as "at war" with unchallenged statistics. Id.

Just as the Supreme Court held in Armstrong, when faced with a similar study, we conclude that the defendants' statistical study offered in this case "did not constitute 'some evidence tending to show the existence of the essential elements of' a selective-prosecution claim," --- U.S. at ----, 116 S. Ct. at 1489 (quoting Berrios, 501 F.2d at 1211).

The district court summarily concluded in this case that Olvis and Palmer satisfied the second prong of the selective-prosecution claim--that the government acted with discriminatory intent--relying solely upon the statistical evidence they presented. The court explained that "[i]nadequately explained evidence of statistical disparity, as presented in this case, is a nonfrivolous showing of discriminatory intent" and found "instructive," but "advisory," a similar type of analysis conducted by the Ninth Circuit in United States v. Armstrong, 48 F.3d 1508 (9th Cir.1995). Unfortunately, when the district court decided this case, it did not have the benefit of the Supreme Court's recent decision rejecting the Ninth Circuit's analysis in Armstrong. --- U.S. at ----, 116 S. Ct. at 1488. In reversing the court of appeals' discovery order, the Supreme Court specifically held that absent an appropriate basis for comparison, statistical evidence alone cannot establish any element of a discrimination claim. Id. at ---- - ----, 116 S. Ct. at 1488-89. Even if the study had a basis for comparison that showed discriminatory effect, it would not necessarily prove discriminatory intent. This is consistent with the general rule that in cases involving discretionary judgments "essential to the criminal justice process," statistical evidence of racial disparity is insufficient to infer that prosecutors in a particular case acted with a discriminatory purpose. See McCleskey, 481 U.S. at 297, 107 S.Ct. at 1769-70 (holding that statistical evidence that race played a role in other death penalty cases was insufficient to create inference that prosecutors sought death penalty against particular defendant because of his race).

In considering the defendants' informal study, the district court also erred in apparently imposing a duty on the government to "explain [the] evidence of statistical disparity" that Olvis and Palmer had presented. The court noted that the defendants' informal study, if left unexplained, demonstrated discriminatory intent. By creating a presumption that unexplained statistical evidence of racial disparity proves racial animus, the court imposed on the government the burden of disproving such animus. As already noted, however, the defendants bear the burden of establishing all elements of their selective-prosecution claim and, to obtain discovery on such a claim, the burden of making a credible showing of "some evidence" on each element. By ruling that defendants can meet these demanding burdens by presenting a study of the type they presented in this case and thereby shifting to the government the onus of dispelling a presumption of discrimination would open virtually every prosecution to a claim for selective prosecution.

On the record presented to the district court, we conclude that Olvis and Palmer have failed to carry their burden of producing some evidence to make a credible showing of both discriminatory effect and discriminatory intent. Accordingly, the district court erred in ordering the United States to provide discovery on its criteria for selecting whom to prosecute. We reverse the district court's dismissal order and remand this case with instructions to reinstate the indictment and for further proceedings.

REVERSED AND REMANDED.