Opinion by Judge REINHARDT;. Concurrence by Judge WALLACE; Dissent by Judge RYMER.
REINHARDT, Circuit Judge:
We review this ease en banc to resolve a conflict in our circuit over the proper standard for determining whether an adequate showing has been made by a defendant seeking discovery in connection with a selective prosecution charge. The conflict arises from two cases filed within days of each other that adopted different approaches to this question. United States v. Redondo-Lemos, 955 F.2d 1296, 1302 (9th Cir.1992), held that the government could be ordered to provide discovery only upon a “prima facie showing that wrongful discrimination is probably taking place.” By contrast, United States v. Bourgeois, 964 F.2d 935, 939 (9th Cir.1992), stated that a prima facie showing was not necessary. Instead, Bourgeois adopted a “color-able basis” test. Id. We conclude that the colorable basis standard better accommodates the competing concerns implicated by discovery in selective prosecution cases.
We have jurisdiction to hear the government’s appeal only because the district judge ordered dismissal of the defendants’ indictments. 18 U.S.C. § 3731. Under 18 U.S.C. § 3731, the government is not permitted to appeal the discovery ruling itself. The statute does, however, permit the government to appeal the dismissal of indictments. Here, the district judge imposed dismissal as a sanction for the government’s failure to comply with her discovery order. That action resulted in an appealable order-under § 3731.
The government does not question the reasonableness of the particular sanction imposed. In fact, it was the government itself that suggested dismissal of the indictments to the district court so that an appeal might lie. On appeal, the government argues only that no sanction at all should have been ordered,' conténding that the district judge abused her discretion in requiring discovery. As a result, the appeal allows us to reach the merits of the underlying discovery issue.
The district judge stayed execution of the dismissal order pending the outcome of this appeal. It appears from the record that she issued the stay so that the defendants would not be released prior to our ruling on the validity of that order. Thus, while our opinion is devoted to a discussion of the discovery order, ultimately we rule on the validity of the order dismissing the indictments.
In sum, the appeal is properly before us only because the government knowingly accepted the consequence of opting for an immediate appeal rather than- complying with the discovery order. That consequence is that, if we affirm, the dismissal of the indictments must now be implemented unless the order dismissing them is further stayed pending review by the Supreme Court. It is too late for the government to change its mind and comply with the discovery order. Were that not the rule, we would simply be permitting appeals of discovery orders under the guise of reviewing dismissal orders that were either only tentative or were never intended to take effect. In either ease, we would not have jurisdiction over the appeals under § 3731.
Because we hold that the defendants here satisfied the colorable basis requirement, we affirm the district court’s dismissal of the indictments.
I.
In April of 1992, defendants Christopher Armstrong, Aaron Hampton, Freddie Mack, Shelton Martin, and Robert Rozelle were charged with federal offenses for their alleged involvement in the distribution of cocaine báse, known colloquially as “crack” or “rock”. The charges stemmed from an in[1511]*1511vestigation conducted under the direction of a joint state and federal task force comprised of detectives from the Inglewood Narcotics Division and agents from the Bureau of Alcohol, Tobacco, and Firearms.
All five defendants were charged with conspiracy to distribute cocaine base under 21 U.S.C. § 846. Some of the defendants were also charged with selling cocaine base under 21 U.S.C. 841(a)(1) and using firearms in connection with drug trafficking in violation of 18 U.S.C. § 924(c). The decision to charge the defendants with federal rather than California state offenses was significant. Federal law imposes a minimum sentence of 10 years and a maximum of life for those convicted of selling more than 50 grams of cocaine base. 21 U.S.C. § 841(b). By contrast, under California law, the minimum sentence for that offense is three years and the maximum is five. Cal.Health & Safety Code § 11351.5 (Deering 1993). All five defendants are black.
On July 20, 1992, defendant Martin filed a Motion for Discovery and/or Dismissal of Indictment for Selective Prosecution. He claimed that the decision to prosecute him on federal charges was based on his race. The other four defendants timely joined the motion which was heard on September 8, 1992.
To support the motion for discovery, the defendants offered into evidence a study of every case involving a charge under 21 U.S.C. §§ 841 and 846 that the Federal Public Defender’s Office for the Central District of California had closed in 1991. The study showed that in all 24 such cases the defendants had been black. At the hearing, counsel for the government responded to the judge’s request for an explanation of these numbers by stating: “I would have no explanation for that. But certainly I can say that there is no racial motivation of any sort that I am aware of as to why we brought this case versus any others.”
The district court granted the motion for discovery. Specifically, the district judge ordered the government to: (1) provide a list of all cases from the prior three years in which *the government charged both cocaine base offenses and firearms offenses; (2) identify the race of the defendants in those cases; (3) identify whether state, federal, or joint law enforcement authorities investigated each ease; and (4) explain the criteria used by the U.S. Attorney’s Office for deciding whether to bring cocaine base cases to the federal court.
The government chose not to comply with the discovery order and instead filed a motion for reconsideration. In support of its motion, the government provided a list of all-defendants charged with violations of 21 U.S.C. §§ 841 and 846 over a three,-year period (without any racial breakdown) as well as declarations by three law enforcement officers and two Assistant United States Attorneys. The declarations collectively provided four explanations for the study’s implication that the overwhelming bulk of federal prosecutions for cocaine base offenses targeted black defendants.
First, the declarations asserted that socioeconomic factors led certain ethnic and racial groups to be particularly involved with the distribution of certain drugs and that blacks were particularly involved in the Los Ange-les-area crack trade. Second, the declarar tions contended that during the three-year period, seven non-black defendants had been prosecuted on federal cocaine base charges, although it appears that all of them were members of racial or ethnic minority groups. (Later, the government identified four more non-black defendants who had been prosecuted during that three-year period, all of whom were also persons of color.) Third, the declarations asserted that many blacks had been tried in state court for cocaine base offenses. Fourth, the declarations contained a description of some of the general factors on which federal prosecutors based their charging decisions for crack-related offenses. The factors specifically referred to. were ' the strength of the evidence, the deterrent value of bringing the charge, the federal interest in the prosecution, and the suspect’s criminal history. The declarations also referred to other unidentified “race-neutral” criteria.
In response, the defendants bolstered the statistical study they had submitted at.the initial hearing with additional declarations. First, one of the defendant’s counsel stated [1512]*1512that'she had spoken with a halfway house intake coordinator who told her that in his experience treating cocaine base addicts, whites and blacks dealt and used the drug in equal numbers. Second, another defense attorney asserted that his experience and conversations with judges, lawyers, and defendants led him to -conclude that many non-blacks were prosecuted for cocaine base offenses in state court. Finally, the defendants submitted an article from the Los An-geles Times discussing the disparate federal sentences imposed for cocaine base and regular cocaine offenses.
District Judge Consuelo Marshall denied the motion for reconsideration. She stated her reasons for the. denial at the hearing: “The statistical data provided by the Defendant raises a question about the motivation of the Government which could be satisfied by the government disclosing its criteria, if there is any criteria, for bringing this case and others like it in Federal court. Without. this criteria the statistical data is evidence and does suggest that the decisions to prosecute in Federal court could be motivated by race. Without expert testimony, this Court cannot conclude that the defendants’ evidence is explained by social phenomena.”
The government again chose not to comply with the discovery order. After some discussion among the parties and the court, the defendants moved to dismiss the indictments as a sanction. The district judge dismissed the indictments, but stayed the order pending appeal. The government timely appealed.
II.
We review for abuse of discretion a district' court’s decision to order discovery. United States v. Bourgeois, 964 F.2d 935, 937 (9th Cir.1992). We have previously stated that, “[t]he task of safeguarding the rights of criminal defendants ultimately rests with the experienced mén and women who preside in our district courts.” United States v. Balough, 820 F.2d 1485, 1491 (9th Cir,1987). District, judges are well situated to' observe possible discrimination in the government’s charging decisions. They have direct experience with the policies and practices of the United States Attorneys in their districts and have the opportunity to discern patterns of discrimination. See United States v. Redondo-Lemos, 955 F.2d 1296, 1298, 1302 (9th Cir.1992).
We are mindful that it is only a discovery order that we are reviewing. To obtain discovery, the defendants need not prove impermissible discrimination. To order discovery, the district court need not decide that the defendants have in fact demonstrated the existence of selective prosecution. If such conclusive determinations could be made without discovery, there would be no need for discovery in the first place. Thus, the evidence necessary to obtain a discovery order when a charge of selective prosecution has been made is obviously substantially less than that needed to prove the charge itself.
We conclude that discovery may be ordered when the evidence provides a color-able basis for believing that discriminatory prosecutorial selections have occurred. The existence of a colorable basis must be judged in light of all the evidence presented to the district court and not simply that offered by the defendant. “The United States Attorney must be given the chance to make whatever showing [he] deems appropriate to dispel the district judge’s concerns.” Redondo-Lemos, 955 F.2d at 1302; see also Bourgeois, 964 F.2d at 941 (considering government’s response). However, if the government fails to make a showing sufficient to dispel those concerns, then a colorable basis remains even though the government has provided some evidence in response.
The colorable basis standard is met by “some evidence tending to show the essential elements of the claim.” United States v. Heidecke, 900 F.2d 1155, 1159 (7th Cir. 1990). Of course, “some evidence” means the showing must be more than frivolous and based on more than conclusory allegations. As we stated in Bourgeois,
to obtain discovery on a selective prosecution claim, a defendant must present specific facts, not mere allegations, which establish a colorable basis for the existence of both discriminatory application of a law [1513]*1513and discriminatory intent on the part of government actors.
964 F.2d at 939.
Because the meaning of “colorable basis” first articulated in Bourgeois has proved elusive, we believe it necessary to explain the standard more fully. In particular, we address here three aspects of the showing needed for discovery in a selective prosecution claim that may be unclear after Bourgeois and concerning which the parties strongly disagree.
First, Bourgeois stated that the colorable basis standard sets a “high threshold” that should rarely justify discovery. Id. at 940. This characterization of the standard appears to conflict with the opinion’s earlier conclusion that the showing needed for discovery is less than that needed for a prima facie case. Id. at 939. In describing the colorable basis test as setting a “high threshold” that would be met only infrequently, our opinion resulted in more confusion than clarification. The “high threshold” language in Bourgeois appeared to set an artificially onerous burden in such claims. The inclusion of that language in our opinion was in error.
Second, Bourgeois did not adequately • explicate what showing is necessary to create a colorable basis for believing that particular prosecutorial conduct has discriminatory effects and is motivated by a discriminatory purpose. To succeed on a claim of selective prosecution the defendant must Show both that the prosecutorial selection “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). However, Wayte made clear that “[i]t is appropriate to judge selective prosecution claims according to ordinary equal protection standards.” Id.
A direct showing of discriminatory intent is not always necessary to make out an equal protection claim; under ordinary equal protection standards, a claimant may prove discriminatory purpose circumstantially. See Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 266, 97 S.Ct. 555, 563-64, 50 L.Ed.2d 450 (1977). A circumstantial showing of intent may be based on evidence of discriminatory effects. “Evidence of the discriminatory impact of decisions is one sort of circumstantial evidence supporting an inference of ... intent.” Diaz v. San Jose Unified School District, 733 F.2d 660, 662 (9th Cir.1984) (en banc), cert. denied 471 U.S. 1065, 105 S.Ct. 2140, 85 L.Ed.2d 497 (1985); see also Batson v. Kentucky, 476 U.S. 79, 93, 106 S.Ct. 1712, 1721, 90 L.Ed.2d 69 (1986) (“Circumstantial evidence of invidious intent may include proof of disproportionate impact.”).
Moreover, we have previously stated that statistical disparities alone may suffice to provide the evidence of discriminatory effect and intent that will establish a prima facie case of selective prosecution. See Redondo-Lemos, 955 F.2d at 1301-02.1 Because the standard for a discovery showing is lower than that for a prima facie case, Bourgeois, 964 F.2d' at 939, we hold that inadequately explained evidence of a significant statistical disparity in the race of those, pros[1514]*1514ecuted suffices to show the colorable basis of discriminatory intent and effect that warrants discovery on a selective prosecution claim.
Finally, Bourgeois did not sufficiently emphasize that judges considering discovery requests on selective prosecution charges should bear in mind the evidentiary obstacles defendants face. The notorious difficulty of proving a race discrimination claim is particularly acute in the context of selective prosecution claims. Cf. Wayte, 470 U.S. at 624, 105 S.Ct. at 1539 (Marshall, J., dissenting). The broad discretion that prosecutors possess over charging decisions means that they alone will often possess the only information ■that would demonstrate such discrimination. Id. As a result, the data necessary to a showing of selective prosecution are far less accessible to. the defendants than to the government.
Defendants attempting to show a colorable basis that warrants discovery can only be expected to make good faith efforts to obtain whatever evidence is readily available, as well as to provide whatever evidence is already in their possession. Cf. Arlington Heights, 429 U.S. at 266, 97 S.Ct. at 563-64 (a court assessing an equal protection claim based on race discrimination must engage in “a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.” (emphasis added)). Defendants are not required to present sophisticated regression analyses closely following the dictates of the scientific method nor a “smoking gun” that irrefutably demonstrates the existence of prosecutorial bias. Nor are defendants required to compile facts which are not easily obtainable by them, such as the racial breakdown and offense characteristics of defendants represented by other counsel.
The colorable basis standard we enunciate today brings our approach to discovery in selective prosecution claims more in line with that of the other circuits. While all the circuits have considered the issue, only four apply a stricter test. Moreover, each of those four requires a prima facie showing,2 a standard which even the dissent rejects as being too high. See Post at 1526-27. By contrast, the “colorable basis” test we set forth mirrors the approach taken by the Third, Sixth, Seventh, Tenth, and D.C. Circuits. Each of those circuits permits discovery when the defendants introduce some evidence tending to show the essential elements of selective prosecution and the government fails to explain it adequately.3 Finally, our approach is certainly more akin to that favored by the two circuits that have adopted a version of the non-fiivolousness standard advocated by the dissenters in Wayte.4
The colorable basis standard not only best reflects the prevailing view of the law, but it also effectively accommodates the twin but conflicting concerns that discovery in selective prosecution claims implicates. On the one hand, selective prosecution claims call into question the very integrity of our system of criminal justice by suggesting that prose-cutorial decisions have been “‘deliberately [1515]*1515based upon an unjustifiable standard such as race, religion, or other arbitrary classification.’ ” Wayte v. United States, 470 U.S. at 608, 105 S.Ct. at 1531 (quoting Bordenkircher v. Hayes, 434 U.S. 357, 364, 98 S.Ct. 663, 668-69, 54 L.Ed.2d 604 (1978)) (further quotation omitted). This concern compels district courts to be vigilant in ensuring that impermissible prosecutorial biases do not remain hidden in files kept from public view.
On the other hand, selective prosecution claims invite the courts to investigate the discretionary charging decisions of executive branch officials that traditionally have not been subject to strict oversight. Separation of powers concerns and the systemic costs that such judicial intrusions entail caution against setting the threshold showing for discovery so low that even frivolous assertions of prosecutorial bias may require the government to lay bare its files. See Wayte,- 470 U.S. at 607, 105 S.Ct. at 1530.
The colorable basis standard ensures that the government will not be called to answer for its charging decisions as a result of frivolous and unwarranted allegations. At the same time, the standard ensures that defendants will not face unjustified hurdles at the discovery stage that will preclude them from demonstrating the existence of actual discrimination in the selection of defendants for criminal prosecution.5
III.
Here, the defendants have presented sufficient evidence to provide a colorable basis for believing that the government has engaged in discriminatory prosecution. They have provided statistical evidence suggesting that blacks are disproportionately charged with federal crack offenses. The government has not offered evidence in response that is sufficiently persuasive to refute the inference that may reasonably be drawn from the statistics. Thus, the district judge is not precluded from determining in her discretion that a colorable basis for selective prosecution has been shown.
A.
The study conducted by the Office of the Federal Public Defender provides a colorable basis for concluding that invidious discrimination may have occurred. The study found that, of all cocaine base cases closed by the Office in 1991, 24 out of 24 involved black defendants. To be sure, such a small number of cases does not conclusively establish either of the elements of selective prosecution. However, the fact that every single crack defendant represented by the Federal Public Defender in all cases that terminated during 1991 was black provides a colorable basis for believing that the challenged prose-cutorial policies are driven by discriminatory motives arid yield discriminatory effects. As a result, the study raises enough of a question to justify further inquiry.
The evidence of discriminatory prosecution in this case is much stronger than the evidence we held insufficient in Bourgeois. In that case, the defendants argued that they were entitled to discovery based on a showing that all prosecutions for firearms violations steinming from a two-day police operation involved black defendants. The district court rejected the defendants’ claims, holding in part that two days was far too short a period to serve as a basis for analyzing the overall conduct of a prosecutorial agency. Instead of focusing on the single operation that resulted in the arrests of Bourgeois and his associates, the district court looked instead to all firearms prosecutions over a two-year span. The court found that the government had prosecuted over 140 people during [1516]*1516that period for the same crimes as those for which Bourgeois was prosecuted. Because Bourgeois “did not allege that all or most of these cases involved blacks,” Bourgeois, 964 F.2d at 940, the district court concluded that he had not established a colorable basis to justify discovery. We held that the district court did not abuse its discretion in refusing to order discovery in these circumstances. See id. at 941.
The evidence offered in Bourgeois involved only one operation, over a single two-day period. The Federal Public Defender study, by contrast, involved an agency that represents a significant percentage of all federal criminal defendants, and it involved all cases closed over a significant period of time. Common sense indicates that such a study provides a much stronger basis for reasonably inferring invidious discrimination than does an analysis of only a single, short police operation. See id. (“[T]he relevant inquiry is the history of prosecutions over a reasonable period of time.”).
The Federal Public Defender study showed that the largest single provider of legal services to federal criminal defendants in the Central District of California did not close a single case involving a crack charge against a non-black defendant in all of 1991. Although the number of cases is too small to resolve the issue either way, it certainly constitutes “some evidence tending to show the essential, elements” of the defendants’ claim.
The district judge based the discovery order on precisely this reasoning. “I think the number is adequate that would at least require the Government to provide some explanation. The time period is such that would require some explanation. The charges are the same or similar, and the race is the same in each case.” A district judge’s decision to permit further inquiry into the issue based on such an analysis cannot fairly be considered to be an abuse of discretion.
B.
In determining whether a color-able basis for discovery exists, the district judge cannot discount the government’s attempts to explain the evidence introduced by the. defendant. A colorable basis must still exist after all the evidence presented by both sides has been considered. However, in evaluating the adequacy of the government’s response, the judge may exercise considerable discretion. We, in turn, give broad deference to the district court’s determinations. See Bourgeois, 964 F.2d .at 937.
Here, the government proffers three responses to the defendants’ showing. First, the government challenges the defendant’s study and contends that its limited scope could not reasonably suggest any race-based disparity in the treatment of crack offenders by federal prosecutors. Second, the government contends that even if the defendants’ showing did suggest such a racial disparity, socioeconomic forces and race-neutral charging criteria rather than prosecutorial discrimination explain that disparity. Finally, the government contends that the defendants made no showing of discriminatory purpose. The district judge did not abuse her discretion in concluding that these responses do not effectively dispel the colorable basis for discrimination that the defendants’ showing created.
The government contends that the study does not demonstrate any discriminatory effect because it shows only that blacks have been prosecuted, not that others of different races and similarly situated have not. But a defendant is not required to demonstrate that the government has failed to prosecute others who are similarly situated. He need only provide a colorable basis for believing that other similarly situated persons have not been prosecuted. The study introduced by defendants clearly satisfies this requirement. “At a threshold level, whether or not there is a significant disparity in the treatment of classes of defendants can normally be determined on the basis of statistical evidence, without reference to the underlying facts of individual cases.” Redondo-Lemos, 955 F.2d at 1301.
Given the prevalence of all kinds of drugs throughout our community, at least some crack distributors are likely to be non-blacks. We must start with the presumption that people of- all races commit all types of crimes — not with the premise that any type [1517]*1517of crime is the exclusive province of any particular racial or ethnic group. These presumptions and premises are, of course, not conclusive. Still, absent some compelling contrary evidence, we must assume that crime knows no exclusive race or creed.6
Here, the government’s own showing in and of itself provides evidence that similarly situated potential defendants of other races do exist — in other words that non-blacks also' commit violations of 21 U.S.C. §§ 841 and 846. Thus, here the government does not take the unsupportable position that only blacks commit the crimes with which the defendants are charged.
The fact that the government has identified some non-black crack offenders who. have been charged with federal crimes, however, does not undermine the accuracy or the force of the study. The government has not identified a single non-black cocaine base case closed by the Federal Public Defender’s Office (or anyone else) during the study’s one-year time period. Instead, the government introduced evidence to show that over a three-year period some federal cocaine base prosecutions of members of other minorities did occur. None of the cases, however, fell within the parameters of the study. As a result, the government’s identification of these cases fails to undermine the accuracy of the study in any respect.7
In the absence of a response undermining the study’s accuracy, the district judge appropriately gave credence to the defendants’ showing that a study of federal cocaine base cases closed by the federal public defender over a particular period of time showed that only black defendants were involved. These results are highly suggestive of the fact that blacks constitute at the very least a disproportionate number of those charged with federal crack offenses.
Of course, the government could have demonstrated that the study’s results were a mere statistical fluke arid that pure chance explains the fact that the only cases closedduring the relevant period were those involving black defendants. But the government failed to do that as well. Instead, the government’s evidence showed only that during a three-year period in which several thousand people were charged with violations of §§ 841(a)(1) and 846, it was only able to identify 11 non-bláck defendants charged with crack cocaine offenses under these statutes, all of whom were members of racial or ethnic minority groups. The identification of a handful of non-black crack offenders (all of whom also happened to be persons of color) does nothing to suggest that the study’s inference of race-based disparity is untenable. Indeed, the fact that the government could identify so few non-blacks charged with federal crack offenses is entirely consistent with the study’s inference that those charged federally with cocaine basé offenses belong virtually exclusively to one racial group.
Of course, in finding that the Federal Public Defender study provides a colorable basis for concluding that discriminatory prosecution may have occurred, we do not determine that the study establishes that the United States Attorney has engaged in such invidious conduct. Rather, it serves only as some evidence that tends to show that the United States Attorney may be engaging in discrimination and suggests thajt further inquiry on the subject is warranted. That is precisely [1518]*1518the course the district court took here by permitting discovery, and it did not abuse its discretion in doing so.
The government argues in the alternative that even if -the defendants’ evidence does suggest a disparity in the treatment of crack offenders on the basis of race, that disparity can be explained by reasons apart from the existence of discriminatory prosecution. Explanations for apparent racial disparities can in some circumstances effectively explain evidence that on its own would provide a basis for ordering discovery. For example, in Bourgeois, discovery was denied largely because the government had offered a compelling explanation of why all those arrested in the two-day sweep were black. The government there explained that the sweep had focused on a particularly dangerous criminal gang of which all members were black. The defendants had made no showing that an overwhelming majority of gang prosecutions focused on black defendants or that the two-day action was in any way representative of what occurred over a longer period. Thus, the government’s response provided the court with a compelling explanation of why all those prosecuted during the sweep were black. See Bourgeois, 964 F.2d at 941.
In its attempt to explain the alleged disparity here, the government offers essentially two responses. Neither response is of a kind that we can hold adequate to dispel the defendants’ showing of a colorable basis. First, the government asserts that blacks in fact commit crack related offenses in disproportionate numbers. As the district judge noted, this assertion is based only on the generalized statements of DEA agents and not expert sociological testimony. Moreover, the response does not speak to the inference that blacks are more frequently prosecuted for federal crack offenses than even their allegedly disproportionate involvement with the drug would justify.
The defendants additional submissions addressed the latter point. On the government’s motion for reconsideration, the defendants introduced two declarations from defense attorneys that provided additional support for the conclusion that the charging decisions may have been discriminatory. One declaration set forth a statement made by an intake coordinator at a Pasadena halfway house who said that, in his experience, there were an equal number of white and non-white users and dealers of crack. The other declaration was from David R. Reed, an experienced criminal defense attorney in the Central District. Reed stated that, in his experience as a federal criminal defense lawyer and as a director of the state court indigent defense panel, he had never handled, known of, or heard of a single federal crack cocaine case involving non-black defendants, but that he knew of many crack cocaine prosecutions against non-blacks in state court.8
The government’s response that blacks are disproportionately involved in the crack trade is simply a non-sequitur. Even if it were true that blacks commit crack offenses in greater numbers' than others, that fact would not explain why black violators are more likely than their non-black counterparts in the crack trade to be charged with a federal than a state offense. In short, the response simply does not undermine the inference from the evidence, including the Reed declaration, that a crack offender’s race is a factor that prosecutors rely on in deciding whether to charge him with a federal offense. Precisely this gap in the government’s explanation caused the district judge to believe discovery was needed. “[Wjhat the court wants to know is whether or not there is any criteria in deciding which of these cases will be filed in state court versus Federal court and if so, what is that criteria. That is the problem I think that needs to be addressed, because we do see a lot of the cases and one does ask why some are in state court and some are being prosecuted in Federal court, and if it’s not based on race, what’s it based on?”
[1519]*1519Second, the government attempted to assure the court that its charging decisions had nothing to do with race. To dispel the inference that race plays a part in the charging decisions of federal prosecutors, the government provided the district court with a general description of some of the criteria it relies on in making such judgments. The government submitted a declaration from David C. Scheper, Assistant United States Attorney and Chief of the Major Crimes unit in the United States Attorney’s Office for the Central District of California, in which he purported to delineate the' criteria. “All charging decisions are made on the basis of whether a federal offense that meets this Office’s guidelines has occurred, the overall strength of the evidence, the deterrence value and federal interest associated with the particular case, the criminal history of the suspects, and other race-neutral criteria.” (emphasis added). It was well within the district court’s discretion to conclude that the recitation by Mr.- Scheper of these vague, generalized, and even unspecified criteria was insufficient to rebut the evidence submitted by the defendants. Just as defendants may not rely on conelusory assertions to meet the colorable basis test, the government may not rely on its alleged use of amorphous criteria to explain a showing that satisfies the standard.9
Given the circumstantial evidence of discrimination that the statistical evidence provides, the government’s simple assertion that it relies on unspecified race-neutral criteria cannot suffice to foreclose further inquiry. More in the way of concrete facts is necessary. Blanket denials of discrimination, though often (but not always) made in good faith, are to be expected in cases such as these. The availability of discovery must not turn on the unlikely event that a federal prosecutor will confess to private biases. As the district judge noted in assessing the government’s denial of racist motivations, “I’m not sure we’d find anybody that would say [otherwise].”
Finally, we note that the government’s contention that no evidence of discriminatory purpose has been shown is unavailing. As our earlier discussion makes clear, an unexplained race-based statistical disparity can itself be sufficient circumstantial evidence of discriminatory intent to justify discovery. Cf. Redondo-Lemos, 955 F.2d at 1301 (finding that such evidence may establish a prima facie case); United States v. Gordon, 817 F.2d 1538, 1541 (11th Cir.), rev’d and vacated in part on other grounds, 836 F.2d 1312 (1988) (finding that circumstantial evidence of intent establishes a prima facie case). We conclude that the Federal Defender Study, analyzing as it did the cases closed over a full year period, provided such circumstantial evidence of intent as is required to be shown at the discovery stage.
C.
When all of the evidence is taken together, the defendants have established several “specific facts, not mere allegations, which establish a colorable basis” to believe that the government has engaged in selective prosecution. Bourgeois, 964'F.2d at 939. Moreover, the government has not explained the evidence of a colorable basis in a manner that would justify our holding that the district judge abused her discretion in concluding that further statistical inquiry and elaboration of governmental charging criteria was justified. A district judge observes countless prosecutions in the course of even' a brief stay on the bench. There is no reason to believe that such judges are easily convinced that the prosecutors who appear before them on a daily basis may be carrying out discriminatory policies. When a district judge does find that there is evidence that reasonably permits one to find a colorable basis for concluding that race-based'prosecutorial se[1520]*1520lection may have occurred, and when the government does not provide an explanation that persuasively undermines such a conclusion, a reviewing court should be reluctant to discount that judgment.
IV.
There are few claims as serious as the charge put forth by the defendants here— that the government has selected them for prosecution because of their race. Such claims deserve the most careful examination by the courts so that the prosecutorial power does not become a license to discriminate based on race. Discovery is the crucial means by which defendants may provide a trial judge with the information needed in order to determine whether a claim of selective prosecution is meritorious. There is no basis for concluding that Judge Marshall abused her discretion in trying to conduct a careful examination of the defendants’ charges or in determining that the defendants made a colorable showing of discriminatory enforcement of the law. Judge Marshall acted properly by authorizing the defendants to inquire further into this issue. Her discovery order was clearly within her discretion. We affirm the sanction she imposed for the government’s failure to comply. As a consequence, the dismissals of the indictments are
AFFIRMED.