WALLACE, Circuit Judge:
The federal government appeals the district court’s order excluding evidence seized by Utah state officials in a trash search operation about which the federal government failed to notify the defense until a few weeks before trial was to begin, even though the state had obtained the evidence two years earlier. It also appeals the district court’s subsequent dismissal of the action with prejudice for failing to proceed to trial before the appeal of the exclusion order was resolved. The district court had jurisdiction over this case pursuant to 18 U.S.C. § 3231. We have jurisdiction pursuant to 18 U.S.C. § 3731. We reverse and remand.
I
Early in 1982, the Sacramento field office of the Federal Bureau of Investigation (FBI) began investigating operations at the Los Gatos, California, office of Sunburst Industries, a company with offices there and in Salt Lake City, Utah. On July 7, 1982, as a result of the Sacramento investigation, a grand jury returned an indictment charging the defendants with 47 counts of mail fraud, wire fraud, interstate transportation of forged or altered securities, and conspiracy. The district court ordered the federal government to provide discovery “in accordance with FRCrimP 16.” Over the next eighteen months, there were numerous pretrial motions dealing with discovery and admissibility of evidence. Finally, a jury trial was set for April 24, 1984.
[1043]*1043On March 28, 1984, however, the federal prosecutors notified defense counsel that they had just discovered, and intended to introduce into evidence during their case-in-chief, numerous documents obtained by the State of Utah’s Organized Crime and Criminal Identification Bureau (Utah Bureau) in a trash search operation conducted at the Salt Lake City office of Sunburst Industries between November 30, 1981 and May 19, 1982. They claimed that a conversation between Sacramento-based FBI agents and a Utah state official on March 21, 1984 had made them aware of the documents for the first time. The district court granted a defense motion for an evidentiary hearing on why the existence of this evidence had not been disclosed earlier in the discovery process.
After holding a six-day evidentiary hearing, the district judge issued an order excluding any evidence the federal government obtained from the Utah authorities which the Utah authorities had seized pursuant to the trash search operation, as well as such evidence obtained as a result of the trash search. He also ordered the government to identify any evidence it possessed and intended to introduce during trial that could be directly traced to the trash search operation. The trial date was continued to May 1, 1984.
The district judge based his ruling on the authority of rule 16(d)(2), Fed.R.Crim.P., and his inherent supervisory powers. He explicitly did not base it on the fourth amendment. He observed that rule 16(a)(1)(C) required the government to make available to the defense any documents within the government’s possession, custody, or control that are material to the defense or intended for use in the government’s case-in-chief. He further stated that rule 16(d)(2) gives federal courts the discretionary authority to grant a continuance, exclude evidence, or enter any other order deemed just in the circumstances if any party fails to comply with a discovery order. The district judge also reasoned that he had inherent supervisory power to punish discovery misconduct, citing United States v. Gonsalves, 691 F.2d 1310, 1315-22 (9th Cir.1982), vacated and remanded, — U.S.-, 104 S.Ct. 54, 78 L.Ed.2d 73 (1983), for the proposition that his supervisory power gives him the authority to preserve the integrity of the judicial system, governed solely by notions of fair play, which are more exacting than minimum requirements of due process.
The district judge concluded that the facts of this case allowed him to exercise both his rule 16(d)(2) and his supervisory powers against the government because the federal prosecutors and the Sacramento FBI agents helping them were either negligent or reckless in not discovering and disclosing the existence of the trash search evidence in a more timely manner. The record shows that neither the Sacramento-based officials nor the Salt Lake City FBI agent who served as a liaison between the Sacramento investigators and the federal investigation of Sunburst’s Utah activities actually possessed, had custody of, or controlled the trash search evidence before March 27, 1984. Moreover, the district judge never made any finding to the contrary. He merely found that the federal officials were negligent or reckless in not learning about, asserting control over, and disclosing the existence of the documents earlier. He found that a number of federal officials unconnected with the Sacramento investigation had at least heard about the trash search operation as early as December 1981. He also found that the Salt Lake City FBI agent had heard about the operation and had passed on one document which could be traced to the trash search operation long before the Sacramento-based officials alleged that they learned about the operation. The district judge said he found it inconceivable that the experienced Salt Lake City FBI agent would not have reported the existence of the trash search material to his fellow investigators in Sacramento. He made no finding, however, that the federal prosecutors had any advance knowledge of the trash search or that the Salt Lake City FBI agent realized the extent or importance of the documents in the state’s possession. Moreover, he [1044]*1044stopped short of finding that the Sacramento-based FBI agents were lying about their asserted ignorance. The district judge concluded, however, that the behavior of the Sacramento-based FBI agents amounted either to negligence or recklessness because not even bureaucratic difficulties could excuse their failure to learn about the trash search.
The district judge reasoned that these facts neither required him to dismiss the case nor allowed him merely to order a continuance, which he admitted was the normal remedy. He stated that dismissal was unacceptable because there was no proof that the government had intentionally or willfully concealed the evidence, the problem occurred before trial began, the evidence was not exculpatory, and any tainted evidence left over could be dealt with at trial on a piece-by-piece basis. He maintained that a continuance was insufficient, however, because it failed to satisfy either the demands of due process or notions of fair play and substantial justice, because the time, effort, and money that would be lost would be prejudicial, unfair, and unjust to the defendants. He observed that the indictment had been filed nearly two years before, and thus the sword had hung over the defendants long enough, that they had spent sufficient money on their defense, and that the court had expended sufficient judicial resources in managing the case. Moreover, the trial was set to begin in four days, and postponing it would inconvenience both the court and the defense attorneys in reclearing their calendars. Finally, a related state trial would be delayed if a continuance were ordered.
The government appealed this exclusionary order pursuant to 18 U.S.C. § 3731 before trial began. While its appeal was pending, the government unsuccessfully attempted either to stay the commencement of trial or to convince the district court that it had lost jurisdiction over the case while the appeal was pending. On the day trial was to begin, the government refused to proceed on the ground that it was entitled first to exercise its statutory right to appeal the exclusionary order under 18 U.S.C. § 3731. It argued that to proceed would take the meaning out of its right to appeal the exclusionary order because the evidence excluded was substantial proof of material facts. The district judge advised the government of his planned action and then dismissed the case pursuant to rule 48(b), Fed.R.Crim.P., with prejudice. He reasoned that the government was willing to proceed before discovering the trash search evidence, apparently concluding that it had sufficient evidence to convict the defendants. He cited several cases to build the proposition that rule 48(b) gave him the discretionary authority to dismiss a case for unnecessary delay either with or without prejudice, whether or not there are constitutional rights at stake. He justified his dismissal of the case with prejudice on grounds that the government’s refusal to proceed was a “flagrant refusal to abide by this Court’s previous orders,” an “unnecessary delay,” and the “rankest abuse of prosecutorial discretion.” The government also appealed this dismissal, and we granted its motion to consolidate its two appeals.
II
The government argues that the district court’s supervisory power did not provide it with authority to exclude the evidence in this case because the government’s late disclosure did not violate the Constitution or any federal statute. It contends that the district court’s reliance on United States v. Gonsalves, 691 F.2d 1310 (9th Cir.1982), vacated and remanded, — U.S.-, 104 S.Ct. 54, 78 L.Ed.2d 73 (1983) (Gonsalves), demonstrates error because' the Supreme Court subsequently vacated that case in light of United States v. Hasting, 461 U.S. 499, 103 S.Ct. 1974, 76 L.Ed.2d 96 (1983) (Hasting). The government argues that the Court in Hasting limited the exercise of supervisory power to instances in which the courts are implementing a remedy for the violation of recognized rights, preserving judicial integrity by ensuring that a jury’s verdict rests on appropriate considerations, or deterring future illegal conduct. Because the govern[1045]*1045ment’s contentions are all legal questions, we review them de novo. United States v. McConney, 728 F.2d 1195, 1201 (9th Cir.) (en banc), cert. denied, — U.S.-, 105 S.Ct. 101, 83 L.Ed.2d 46 (1984) (McConney).
The Supreme Court first used the label “supervisory power” in McNabb v. United States, 318 U.S. 332, 341, 63 S.Ct. 608, 613, 87 L.Ed. 819 (1943), to describe its authority to formulate procedural rules in its administration of criminal justice. Although federal courts had formulated common-law procedural rules before McNabb, they had not previously claimed any general power to do so, but had instead been careful to identify specific congressional mandates for such rule creation and to abide by any express or implied limits placed by Congress on its mandates. See, e.g., United States v. Reid, 53 U.S. (12 How.) 361, 365, 13 L.Ed. 1023 (1851); Beale, Reconsidering Supervisory Power in Criminal Cases: Constitutional and Statutory Limits on the Authority of the Federal Courts, 84 Colum.L.Rev. 1433, 1438 & n. 28 (1984).
Although the Supreme Court continues to recognize that lower federal courts enjoy supervisory power, it has circumscribed this power to the formulation of “procedural rules not specifically required by the Constitution or the Congress.” Hasting, 461 U.S. at 505, 103 S.Ct. at 1978 (emphasis added). In addition, it has never identified its source. Beale, supra, at 1434, 1462. It is clear, however, that the power has limits. Hasting, 461 U.S. at 505, 103 S.Ct. at 1978.
We asserted in Gonsalves that the source of the power lies in the judiciary’s role in our constitutional system of checks and balances as limited by the doctrine of separation of powers, 691 F.2d at 1318, and that it is necessary to the courts’ role in “cheek[ing] ... governmental excesses of the executive and the legislature” that threaten the courts’ “own institutional integrity.” Id. at 1319. We also claimed that the supervisory power was animated by “notions of fair play which are more exacting than the minimum constitutional requirements of due process,” id. at 1316, and that it allowed federal courts “to do justice.” Id. The district judge relied upon this case in exercising his supervisory power.
But the Supreme Court vacated and remanded our decision in Gonsalves in light of Hasting. This suggests that our extensive discussion was insufficiently sensitive to the power’s limitations. In particular, the Supreme Court’s action calls into question the expansive supervisory discretion we attempted to derive from the supposed animating force of “notions of fair play.” The Court’s emphasis that supervisory power is a power limited to “formulate procedural rules,” 461 U.S. at 505, 103 S.Ct. at 1978, is directly opposed to our contention that the power allows federal courts “to do justice in particular fact situations that do not lend themselves to rules of general application.” Gonsalves, 691 F.2d at 1316.
One prominent theory, but different from our notions in Gonsalves, attempts to justify the exercise of supervisory power as a form of specialized and limited federal common law. See Monaghan, The Supreme Court, 197k Term — Forward: Constitutional Common Law, 89 Harv.L.Rev. 1, 34-38 (1975); cf. Palermo v. United States, 360 U.S. 343, 345, 79 S.Ct. 1217, 1221, 3 L.Ed.2d 1287 (1959) (Palermo). No one seriously disputes that Congress has the authority to create procedural rules to govern criminal proceedings. See Livingston v. Story, 34 U.S. (9 Pet.) 632, 656, 9 L.Ed. 255 (1835); Wayman v. Southard, 23 U.S. (10 Wheat.) 1, 21-22, 6 L.Ed. 253 (1825). Moreover, the power of federal courts to decide cases is at least potentially coextensive with Congress’s legislative power. See Osborn v. Bank of the United States, 22 U.S. (9 Wheat.) 738, 818-19, 6 L.Ed. 204 (1824). Thus, the theory goes, the federal courts may sometimes be justified in creating procedural rules, subject to statutory and separation-of-powers limitations, when injustices arise for which Congress almost certainly would provide a remedy were it feasible for it to do so.
[1046]*1046Congressional success in limiting supervisory rules reinforces the theory that the federal judiciary’s supervisory power is a power it enjoys only concurrently with Congress, and over which Congress has the final say. Congress generally has acquiesced in the supervisory power decisions, but it has been motivated to limit or reverse the Supreme Court’s exercise of this power on at least two occasions. See Jencks Act, Pub.L. No. 85-269, 71 Stat. 595 (1957) (codified as amended at 18 U.S.C. § 3500) (limiting the Court's exercise of supervisory power in Jencks v. United States, 353 U.S. 657, 77 S.Ct. 1007, 1 L.Ed.2d 1103 (1957), which required disclosure of memoranda prepared by government informants); Omnibus Crime Control and Safe Streets Act of 1968, Pub.L. No. 90-351, § 701(a), 82 Stat. 197, 210 (codified as amended at 18 U.S.C. § 3501) (reversing McNabb-Mallory rule). The Supreme Court upheld the Jencks Act in Palermo, 360 U.S. at 345-48, 355-56, 79 S.Ct. at 1221-22, 1226, but it has had no occasion to pass on the validity of the Omnibus Act. See United States v. Manuel, 706 F.2d 908, 912-13 (9th Cir.1983) (recognizing validity of Omnibus Act).
Regardless of whether the supervisory power stems from the federal courts’ inherent power to check intrusions by other branches of government or whether it is a form of specialized federal common law, the separation-of-powers principle imposes significant limits on it. As a threshold matter, a court may not exercise any supervisory power absent “a clear basis in fact and law for doing so.” United States v. Chanen, 549 F.2d 1306, 1313 (9th Cir.), cert. denied, 434 U.S. 825, 98 S.Ct. 72, 54 L.Ed.2d 83 (1977); see also United States v. McClintock, 748 F.2d 1278, 1283-86 (9th Cir.1984). Proper regard for judicial integrity does not justify a “ ‘chancellor’s foot’ veto” over activities of coequal branches of government. United States v. Russell, 411 U.S. 423, 435, 93 S.Ct. 1637, 1644, 36 L.Ed.2d 366 (1973). Judicial integrity is rarely threatened significantly when executive action does not violate the Constitution, a federal statute, or a procedural rule. A federal court that imposes sanctions on executive conduct that is otherwise permitted by the Constitution, a federal statute or a rule will most likely be invading the executive sphere rather than protecting itself from invasion. Similarly, the separation-of-powers principle suggests that the creation of a rule by supervisory power can be justified only when a recognized right has been violated. Even then, it may be wise for the courts to await congressional action.
The Supreme Court has also indicated that the supervisory power may not be used to impose remedies for the vindication of nonconstitutional rights that the Court deliberately has limited in the context of vindicating constitutional rights. See United States v. Payner, 447 U.S. 727, 735-36, 100 S.Ct. 2439, 2446, 65 L.Ed.2d 468 (1980). Thus, for example, courts may not impose an exclusionary rule in the name of exercising their supervisory power to remedy a nonconstitutional violation in seizing evidence to the extent the Supreme Court limits that remedy to fourth amendment violations. Id. at 736, 100 S.Ct. at 2446.
The delay in disclosure in this case did not violate any constitutional provision, federal statute, specific discovery order, or any other recognized right except perhaps rule 16, Fed.R.Crim.P. See United States v. Roybal, 566 F.2d 1109, 1110-11 (9th Cir.1977) (recognizing supervisory power to remedy prejudice of untimely disclosures of evidence in violation of specific discovery order by sanctions as severe as conviction reversals). Moreover, the fact that rule 16 contains specific remedies for its violation eliminates any justification for an exercise of supervisory power to create any other remedy for it.- We hold, therefore, that the separation-of-powers principle barred the district court from exercising any supervisory power to exclude the evidence in this case.
Ill
The government also argues that the district court erred in excluding the evidence [1047]*1047pursuant to rule 16(d)(2), Fed.R.Crim.P., because rule 16 applies only to documents that are within the federal government’s actual possession, custody, or control, because the prosecution’s failure to learn about the trash search documents and to make its disclosure earlier was not negligent or reckless, because a continuance was the appropriate remedy if it did violate rule 16, and because the exclusionary order was overbroad. We agree with the government’s first reason, and therefore need not reach the remaining issues.
Rule 16(d)(2) states that a district court has discretionary authority to “order [a] party to permit the discovery or inspection, grant a continuance, or prohibit the party from introducing evidence not disclosed, or it may enter such other order as it deems just under the circumstances” to remedy a failure by the government to permit inspection and copying of tangible evidence described by rule 16(a)(1)(C), which is evidence “within the possession, custody or control of the government, and which [is] material to the preparation of [the] defense or [is] intended for use by the government as evidence in chief at the trial.” The government intended to use the trash search evidence in its case-in-chief. Thus, the only question is whether the evidence was “within the possession, custody or control” of the federal government.
The record shows that neither the Sacramento-based officials nor the Salt Lake City FBI agent actually possessed, had custody of, or controlled the trash search evidence before March 27, 1984. Moreover,. the district judge never made any finding to the contrary.1 Thus, the precise issue we face is whether rule 16(a)(1)(C) ever requires the federal government to disclose and produce documents that are in the actual possession, custody or control of state officials, the relevance of which the federal government negligently or recklessly fails to appreciate. This is a legal question, re viewable de novo. McConney, 728 F.2d at 1201.
Our first question is whether, pursuant to rule 16(a)(1)(C), the “government” includes any persons other than the prosecutors. We have not faced this question before. The district court appears to have taken the position that the “government” included the Sacramento-based FBI agents assisting the federal prosecutor in this case as well as the FBI agent in Salt Lake City who was coordinating information.
In United States v. Trevino, 556 F.2d 1265, 1271-72 (5th Cir.1977), the court addressed the question of whether a presentence report in the possession of the probation department was in the possession of the “government” for purposes of rule 16(a)(1)(C). While holding that it was not, that court concluded the “government” meant the prosecution, which is “in the business of introducing evidence in chief at trial.” Id. at 1271. In dictum, the court included in that term the prosecutor and closely connected investigative agencies. Id. at 1272. We need not resolve whether the district court correctly included the identified FBI agents for purposes of rule 16(a)(1)(C). Even if we assume the district judge was correct, the critical issue pertains to possession of the records.
The important question, therefore, is whether the documents must be in the actual possession or control of the FBI agents or whether constructive possession or control is sufficient. Under rule 16(a)(1)(A), which deals with statements of a defendant, we have recognized that the prosecutor’s actual possession is not necessary in all cases. For example, in United States v. Bailleaux, 685 F.2d 1105, 1113-14 (9th Cir.1982), we held that the rule imposed a duty on the prosecutor to produce copies of a defendant’s statements that he did not actually possess if the FBI possessed them and the prosecutor could have learned about them by exercising due dil[1048]*1048igence. Similarly, in United States v. Gillings, 568 F.2d 1307, 1310 (9th Cir.) (per curiam), cert. denied, 436 U.S. 919, 98 S.Ct. 2267, 56 L.Ed.2d 760 (1978), we held that a prosecutor breached his duty to produce more timely a tape containing the defendant’s statements that he should have obtained earlier from the Internal Revenue Service.
The cases discussing a prosecutor’s due diligence obligation under rule 16(a)(1)(A) are not relevant to our case under rule 16(a)(1)(C). We have already assumed for purposes of this appeal that the government included the designated FBI agents. Rule 16(a)(1)(A) only requires the government to disclose any of the defendant’s statements that are “within the possession, custody or control of the government, the existence of which is known, or by the exercise of due diligence may become known, to the attorney for the government.” Thus, the due diligence requirement establishing constructive possession relates solely to the prosecutor and whether he should have been aware of a statement in the possession of another federal agency. As no such language is found in rule 16(a)(1)(C), a literal reading of the entire rule requires us to conclude that Congress intended no such constructive possession extension. Moreover, even if such language were found in rule 16(a)(1)(C), it would only create a due diligence requirement over documents in the possession, custody, or control of some federal agency. We would still be required to find some special reason to justify extending the requirement to documents in the possession, custody, or control of state authorities.
Because we find no due diligence language in rule 16(a)(1)(C) at all, nor any special reason to deviate from its plain language, we conclude that it triggers the government's disclosure obligation only with respect to documents within the federal government’s actual possession, custody, or control. This interpretation is consistent with the only case of ours addressing this problem. We have specifically held that rule 16(a)(1)(C) imposes no duty on the federal government to produce a document that is in the actual possession, custody or control of a foreign government, even if the federal government knew about it. United States v. Friedman, 593 F.2d 109, 119-20 (9th Cir.1979). We stated that “[njeither Rule 16 nor § 3500 required production of [the document], because the Government did not have possession or control of it.” Id. at 120. See also United States v. Higginbotham, 539 F.2d 17, 21 (9th Cir.1976) (no duty imposed on federal government to preserve and produce incriminating photographic identification materials that were in state’s exclusive possession until destroyed if state was not acting on federal government’s behalf); Beavers v. United States, 351 F.2d 507, 509 (9th Cir.1965) (Jencks Act does not require federal government to disclose recording of witness statements that is in actual possession of state rather than federal authorities). Cf. United States v. Cadet, 727 F.2d 1453, 1466 (9th Cir.1984) (because defense failed to show documents in government’s possession would be helpful, it could not secure them from either the government or holder pursuant to rule 16(a)(1)(C)); United States v. Armstrong, 621 F.2d 951, 954 (9th Cir.1980) (although rule 16(a)(1)(C) “provides for the inspection and photographing of buildings or places which are within the possession, custody or control of the Government, there are no comparable provisions allowing inspection of the property of third parties”).
The Fifth Circuit has also addressed this issue and concluded that rule 16(a)(1)(C) only applies to documents in the federal government’s actual possession, custody or control. In United States v. Adkins, 741 F.2d 744 (5th Cir.1984), the prosecutor informed Adkins’s counsel before trial that he had relevant bank records that were available for inspection, but that he did not intend to offer them in evidence. Id. at 746. After trial began, however, the prosecutor said he had changed his mind and would introduce the records. Id. He provided copies to the defense, and the court allowed Adkins to defer cross-examining a certain witness for four days in order to [1049]*1049provide Adkins time to examine the records. Id. at 746-47. On appeal, Adkins argued that he was denied a fair trial because the government had lied about its intent before trial and because the records it introduced and provided were incomplete. Id. at 747. The court rejected these arguments on the ground that the government had merely changed its mind and that rule 16(a)(1)(C) only requires disclosure of documents in the government’s possession that it intends to introduce. Id. It also rejected the incompleteness argument on the ground that the records produced were all that were in the government’s possession and rule 16(a)(1)(C) “require[s] the government only to turn over those records actually in its possession.” Id. (emphasis added).
Similarly, in Thor v. United States, 574 F.2d 215 (5th Cir.1978), the defendant attempted to subpoena an address book allegedly in the possession of state officials that he said would exculpate him. Id. at 219. In affirming the district court’s refusal to issue the subpoena, the court of appeals held, in reference to rule 16(a)(1)(C), that the federal government need not produce documents solely in the control of county officers. Id. at 220-21.
Therefore, we conclude that the triggering requirement under rule 16(a)(1)(C) is that the papers, documents, and tangible objects be in the actual possession, custody or control of the government. Here, they were not. The fashioning of a due diligence constructive possession or control addition to the rule by the district judge was erroneous and, therefore, his suppression of the trash cover evidence based upon rule 16 must be reversed.
The defendants may have been put into a difficult position because the government announced, four weeks before trial, that it had 382 tangible items it planned to introduce into evidence. Although the district court could not suppress the evidence pursuant to any supervisory powers or rule 16, and the parties have cited to us no other authority for such an order, the district judge could have considered more searchingly whether a continuance was appropriate under the circumstances simply as a matter of proper case management. The record is barren as to how much time longer than four weeks, if any, the defendants would have needed to prepare to meet this new evidence — indeed, the defendants did not even request a continuance. Thus, we cannot review whether a continuance would have been appropriate here.
IV
Our holding that the district court erroneously suppressed the evidence will not result in a reversal, however, if the district court properly dismissed the indictment with prejudice for failure to prosecute. The government contends that its initial appeal of the exclusionary order deprived the district court of jurisdiction over the case or, in the alternative, that the court abused its discretion in dismissing the indictment with prejudice when the government’s failure to prosecute was its only way to assure receipt of the full benefit of its right to appeal the exclusionary order.
We reject the government’s argument that the district court lost jurisdiction over the action when the government filed its notice of appeal pursuant to 18 U.S.C. § 3731 to challenge the exclusionary order. In the usual circumstance, “[t]he filing of a notice of appeal ... confers jurisdiction on the court of appeals and divests the district court of its control over those aspects of the case involved in the appeal.” Griggs v. Provident Consumer Discount Co., 459 U.S. 56, 58, 103 S.Ct. 400, 402, 74 L.Ed.2d 225 (1982) (per curiam). Section 3731 appeals, however, are not usual. As we previously observed, a district court retains “the naked power, in appropriate cases, to dismiss an indictment” while a section 3731 appeal from a pretrial order is pending. United States v. Emens, 565 F.2d 1142, 1144 (9th Cir.1977), citing United States v. Cox, 475 F.2d 837, 841 (9th Cir.1973) (upholding dismissal of indictment pending appeal of suppression order). We believe this to be a sound policy.
[1050]*1050The government has a conditional right to appeal a suppression order, but the exercise of this right may result in a disruptive effect on the criminal trial process, therefore harboring a potential for abuse. As a result, the government's right to appeal pretrial suppression orders must be balanced with a defendant’s right to proceed to trial on the indictment. This can best be accomplished, as we stated in Emens, by retaining jurisdiction in the district court to dismiss the indictment in appropriate cases.
In protecting its right to appeal, the government may request the district court to stay proceedings pending appeal and, if refused, pursue the issue with this court. If the indictment is dismissed, as in this case, we review the district court’s order based upon whether there was an abuse of discretion. Cf. United States v. Griffin, 617 F.2d 1342, 1347 (9th Cir.), cert. denied, 449 U.S. 863, 101 S.Ct. 167, 66 L.Ed.2d 80 (1980).
In the case before us, we do not reach the issue whether the district court could have dismissed the indictment without prejudice for unnecessary delay, but only whether it abused its discretion in dismissing the indictment with prejudice. Rule 48(b), Fed.R.Crim.P., provides federal courts with discretionary authority to dismiss an indictment if there is “unnecessary delay in bringing a defendant to trial.” We have recognized that this rule allows a district court to dismiss a case with or without prejudice even when no constitutional rights are at stake. United States v. Loud Hawk, 628 F.2d 1139, 1150 (9th Cir.1979) (en banc), cert. denied, 445 U.S. 917, 100 S.Ct. 1279, 63 L.Ed.2d 602 (1980) (Loud Hawk). But the district court must exercise this authority “with caution,” especially in dismissing a case with prejudice, and it may dismiss with prejudice “only after a forewarning to the prosecution that dismissal with prejudice will result from a failure to proceed to trial.” Id.
Although the district court gave the required warning before dismissing with prejudice, our decision in Loud Hawk shows that the district court failed to exercise sufficient caution. In Loud Hawk, the district court had issued an order excluding evidence relating to dynamite counts in an indictment charging the defendants with several dynamite and nondynamite related violations of federal firearms laws. Id. at 1149-50. The government appealed the order pursuant to 18 U.S.C. § 3731 before trial began, and then refused to proceed to trial. 628 F.2d at 1150. The district court dismissed the entire indictment with prejudice. Id. We reversed the dismissal as an abuse of discretion. We held that the delay was necessary on the dynamite counts to protect the meaning of the government’s appeal pursuant to 18 U.S.C. § 3731 because “the evidence suppressed [was] substantial proof of a fact material in the proceeding.” Id.
In this case, the district court concluded that the indictment should be dismissed with prejudice because of the government’s failure to abide by orders of the court, unnecessary delay, and abuse of prosecutorial discretion. The only order that the court accused the government of violating, however, was the discovery order pursuant to rule 16. As held earlier, there was no such violation. The delay alleged was necessary to appeal the exclusion order. The district judge believed the government could have proceeded with the evidence it possessed before its knowledge of the trash search. That, however, misses the point. The necessity requirement, as we said in Loud Hawk, goes to whether the suppressed evidence constitutes “substantial proof of a fact material in the proceeding.” 628 F.2d at 1150. There was no finding here that it was not, nor would the record support such a finding. If, as we assume, the charged prosecutorial abuse relates to the failure to proceed until the appeal of the suppression order, the record fails to support the ordered dismissal with prejudice. Here, the government had a right to appeal under Loud Hawk. The case is too similar to the action of the prosecutor of the dynamite counts in that case for us to distinguish the two.
[1051]*1051Thus, we conclude that the district judge used the wrong ingredients in striking the necessary balance and abused his discretion in dismissing the indictment with prejudice. The district judge was properly concerned about the situation which confronted the defendants and the court, but in this instance, and based upon this record, the remedy imposed was not available.
REVERSED AND REMANDED.