delivered the opinion of the Court.
The three petitioners were convicted on Count 1 of an indictment brought under 18 U. S. C. § 3711 for conspiracy to defraud the United States with reference to certain tax matters. Petitioner Halperin was also convicted on Counts 5, 6, and 7 of the same indictment, charging him with violating 18 U. S. C. § 15032 by endeavoring corruptly to influence certain witnesses before a grand jury which was investigating matters involved in the conspiracy charged in Count 1 of the indictment. Each petitioner was sentenced to five years’ imprisonment and fined under Count 1. On each of Counts 5, 6, and 7, Halperin was sentenced to two years’ imprisonment and a fine of $1,000, the prison sentences on these Counts and that on Count 1 to run concurrently. The Court of Appeals for the Second Circuit affirmed, with the late Judge Frank dissenting. 233 F. 2d 556. We granted certiorari, 352 U. S. 866, in order to resolve important questions relating to (a) the statute of limitations in conspiracy *394prosecutions, as to which the decision below was alleged to be in conflict with this Court’s decisions in Krulewitch v. United States, 336 U. S. 440, and Lutwak v. United States, 344 U. S. 604; and (b) the use on Halperin’s cross-examination of his prior claim of the Eifth Amendment’s privilege against self-incrimination before a grand jury. For the reasons discussed hereafter, we conclude that these convictions must be reversed, and the petitioners granted a new trial.
On October 25, 1954, a grand jury returned an indictment, Count 1 of which charged petitioners and others with conspiring among themselves and with others “to defraud the United States in the exercise of its governmental functions of administering the internal revenue laws and of detecting and prosecuting violations of the internal revenue laws free from bribery, unlawful impairment, obstruction, improper influence, dishonesty, fraud and corruption . . . .” The indictment further charged that a part of the conspiracy was an agreement to conceal the acts of the conspirators.3 Overt acts within three years of the date of the indictment were charged. Counts 5, 6, and 7 of the indictment charged petitioners with violating 18 U. S. C. § .1503 in the manner already indicated.
The proofs at the trial presented a sordid picture of a ring engaged in the business of “fixing” tax fraud cases *395by the use of bribes and improper influence. In general outline, the petitioners’ scheme, which is set forth in more detail in the Court of Appeals’ opinion,4 was as follows:
In 1947 and 1948 two New York business firms, Patullo Modes and Gotham Beef Co., were under investigation by the Bureau of Internal Revenue for suspected fraudulent tax evasion. Through intermediaries, both firms established contact with Halperin, a New York attorney, and his associates in law practice. Halperin in turn conducted negotiations on behalf of these firms with Grüne-wald, an “influential” friend in Washington, and reported that Grünewald, for a large cash fee, would undertake to prevent criminal prosecution of the taxpayers. Grüne-wald then used his influence with Bolich, an official in the Bureau, to obtain “no prosecution” rulings5 in the two tax cases. These rulings were handed down in 1948 and 1949. Grünewald, through Halperin, was subsequently paid $60,000 by Gotham and $100,000 by Patullo.6
Subsequent activities of the conspirators were directed at concealing the irregularities in the disposition of the Patullo and Gotham cases. Bolich attempted to have the Bureau of Internal Revenue report on the Patullo case “doctored,” and careful steps were taken to cover up the traces of the cash fees paid to Grünewald. In 1951 a congressional investigation' was started by the King Committee of the House of Representatives; the conspirators felt themselves threatened and took steps to hide their traces. Thus Bolich caused the disappearance *396of certain records linking him to Grünewald, and the taxpayers were repeatedly warned to keep quiet. In 1952 the taxpayers and the conspirators were called before a Brooklyn grand jury. Halperin attempted to induce the taxpayers not to reveal the conspiracy, and Grünewald asked his secretary not to talk to the grand jury. These attempts at concealment were, however, in vain. The taxpayers and some of Halperin’s associates revealed the entire scheme, and petitioners’ indictment and conviction followed.7
The first question before us is whether the prosecution of these petitioners on Count 1 of the indictment was barred by the applicable three-year statute of limitations.8
The indictment in these cases was returned on October 25, 1954. It was therefore incumbent on the Government to prove that the conspiracy, as contemplated in the agreement as finally formulated, was still in existence on October 25, 1951, and that at least one overt act in furtherance of the conspiracy was performed after that date.9 For where substantiation of a conspiracy charge *397requires proof of an overt act, it must be shown both that the conspiracy still subsisted within the three years prior to the return of the indictment, and that at least one overt act in furtherance of the conspiratorial agreement was performed within that period. Hence, in both of these aspects, the crucial question in determining whether the statute of limitations has run is the scope of the conspiratorial agreement, for it is that which determines both the duration of the conspiracy, and whether the act relied on as an overt act may properly be regarded as in furtherance of the conspiracy.10
Petitioners, in contending that this prosecution was barred by limitations, state that the object of the conspiratorial agreement was a narrow one: to obtain “no prosecution” rulings in the two tax cases. When these rulings were obtained, in October 1948 in the case of Gotham Beef, and in January 1949 in the case of Patullo Modes, the criminal object of the conspiracy, petitioners say, was attained and the conspirators’ function ended. They argue, therefore, that the .statute of limitations started running no later than January 1949, and that the *398prosecution was therefore barred by 1954, when the indictment was returned.11
The Government counters with two principal contentions: First, it urges that even if the main object of the conspiracy was to obtain decisions from the Bureau of Internal Revenue not to institute criminal tax prosecutions — decisions obtained in 1948 and 1949 — the indictment alleged,12 and the proofs showed, that the conspiracy also included as a subsidiary element an agreement to conceal the conspiracy to “fix” these tax cases, to the end that the conspirators would escape detection and punishment for their crime. Says the Government, “from the very nature of the conspiracy . . . there had to be, and was, from the outset a conscious, deliberate, agreement to conceal . . . each and every aspect of the conspiracy . . . .” It is then argued that since the alleged conspiracy to conceal clearly continued long after the main criminal purpose of the conspiracy was accomplished, and since overt acts in furtherance of the agreement to conceal were performed well within the indictment period, the prosecution was timely.
Second, and alternatively, the Government contends that the central aim of the conspiracy was to obtain *399for these taxpayers, not merely a “no prosecution” ruling, but absolute immunity from tax prosecution; in other words, that the objectives of the conspiracy were not attained until 1952, when the statute of limitations ran on the tax cases which these petitioners undertook to “fix.” The argument then is that since the conspiracy did not end until 1952, and since the 1949-1952 acts of concealment may be regarded as, at least in part, in furtherance of the objective of the conspirators to immunize the taxpayers from tax prosecution, the indictment was timely.
For reasons hereafter given, we hold that the Government’s first contention must be rejected, and that as to its second, which the Court of Appeals accepted, a new trial must be ordered.
I.
We think that the Government’s first theory — that an agreement to conceal a conspiracy can, on facts such as these, be deemed part of the conspiracy and can extend its duration for the purposes of the statute of limitations— has already been rejected by this Court in Krulewitch v. United States, 336 U. S. 440, and in Lutwak v. United States, 344 U. S. 604.
In Krulewitch the question before the Court was whether certain hearsay declarations could be introduced against one of the conspirators. The declarations in question were made by one named in the indictment as a co-conspirator after the main object of the conspiracy (transporting a woman to Florida for immoral purposes) had been accomplished. The Government argued that the conspiracy was not ended, however, since it included an implied subsidiary conspiracy to conceal the crime after its commission, and that the declarations were therefore still in furtherance of the conspiracy and binding on *400co-conspirators. This Court rejected the Government’s argument. It then stated:
“Conspirators about' to commit crimes always expressly or implicitly agree to collaborate with each other to conceal facts in order to prevent detection, conviction and punishment. Thus the [Government’s] argument is that even after the central criminal objectives of a conspiracy have succeeded or failed, an implicit subsidiary phase of the conspiracy always survives, the phase which has concealment as its sole objective.
“We cannot accept the Government’s contention. . . . The rule contended for by the Government could have far-reaching results. For under this rule plausible arguments could generally be made in conspiracy cases that most out-of-court statements offered in evidence tended to shield co-conspirators. We are not persuaded to adopt the Government’s implicit conspiracy theory which in all criminal conspiracy cases would create automatically a further breach of the general rule against the admission of hearsay evidence.” 13
Mr. Justice Jackson, concurring, added:
“I suppose no person planning a crime would accept as a collaborator one on whom he thought he could not rely for help if he were caught, but I doubt that this fact warrants an inference of conspiracy for that purpose. . . .
“It is difficult to see any logical limit to the ‘implied conspiracy,’ either as to duration or means .... On the theory that the law will impute to the confederates a continuing conspiracy to defeat justice, one conceivably could be bound by *401another’s unauthorized and unknown commission of perjury, bribery of a juror or witness, [etc.] ....
“Moreover, the assumption of an indefinitely continuing offense would result in an indeterminate extension of the statute of limitations. If the law implies an agreement to cooperate in defeating prosecution, it must imply that it continues as long as prosecution is a possibility, and prosecution is a possibility as long as the conspiracy to defeat it is implied to continue.” 14
The Krulewitch case was reaffirmed in Lutwak v. United States, supra. Here again the question was the admissibility of hearsay declarations of co-conspirators after the main purpose of the conspiracy had been accomplished; again the Government attempted to extend the life of the conspiracy by an alleged subsidiary conspiracy to conceal. Although in Lutwak, unlike in Krulewitch, the existence of a subsidiary conspiracy to conceal was charged in the indictment, the Court again rejected the Government’s theory, holding that no such agreement to conceal had been proved or could be implied.
The Government urges us to distinguish Krulewitch and Lutwak on the ground that in those cases the attempt was to imply a conspiracy to conceal from the mere fact that the main conspiracy was kept secret and that overt acts of concealment occurred. In contrast, says the Government, here there was an actual agreement to conceal the conspirators, which was charged and proved to be an express part of the initial conspiracy itself.
We are unable to agree with the Government that, on this record, the cases before us can be distinguished on such a basis.
The crucial teaching of Krulewitch and Lutwak is that after the central criminal purposes of a conspiracy have *402been attained, a subsidiary conspiracy to conceal may not be implied from circumstantial evidence showing merely that the conspiracy was kept a secret and that the conspirators took care to cover up their crime in order to escape detection and punishment. As was there stated, allowing such a conspiracy to conceal to be inferred or implied from mere overt acts of concealment would result in a great widening of the scope of conspiracy prosecutions, since it would extend the life of a conspiracy indefinitely. Acts of covering up, even though done in the context of a mutually understood need for secrecy, cannot themselves constitute proof that concealment of the crime after its commission was part of the initial agreement among the conspirators. For every conspiracy is by its very nature secret; a case can hardly be supposed where men concert together for crime and advertise their purpose to the world. And again, every conspiracy will inevitably be followed by actions taken to cover the conspirators’ traces. Sanctioning the Government’s theory would for all practical purposes wipe out the statute of limitations in conspiracy cases, as well as extend indefinitely the time within which hearsay declarations will bind co-conspirators.
A reading of the record before us reveals that on the facts of this case the distinction between “actual” and “implied” conspiracies to conceal, as urged upon us by the Government, is no more than a verbal tour de force. True, in both Krulewitch and Lutwak there is language in the opinions stressing the fact that only an implied agreement to conceal was relied on.15 Yet when we look to the facts of the present cases, we see that the evidence from which the Government here asks us to deduce an “actual” agreement to conceal reveals nothing beyond that adduced in prior cases. What is this evidence? *403First, we have the fact that from the beginning the conspirators insisted on secrecy. Thus the identities of Grünewald and Bolich were sedulously kept from the taxpayers; careful steps were taken to hide the conspiracy from an independent law firm which was also working on Patullo’s tax problems; and the taxpayers were told to make sure that their books did not reflect the large cash payments made to Grünewald. Secondly, after the “no prosecution” rulings were obtained, we have facts showing that this secrecy was still maintained. Thus, a deliberate attempt was made to make the above-mentioned independent law firm believe that it was its (quite legitimate) efforts which produced the successful ruling. Finally, we have the fact that great efforts were made to conceal the conspiracy when the danger of exposure appeared. For example, Bolich got rid of certain records showing that he had used Grunewald’s hotel suite in Washington; Patullo’s accountant was persuaded to lie to the grand jury concerning a check made out to an associate of the conspirators; Grünewald attempted to persuade his secretary not to talk to the grand jury; and the taxpayers were repeatedly told by Halperin and his associates to keep quiet.
We find in all this nothing more than what was involved in Krulewitch, that is, (1) a criminal conspiracy which is carried out in secrecy; (2) a continuation of the secrecy after the accomplishment of the crime; and (3) desperate attempts to cover up after the crime begins to come to fight; and so we cannot agree that this case does not fall within the ban of those prior opinions.
In effect, the differentiation pressed upon us by the Government is one of words rather than of substance. In Krulewitch it was urged that a continuing agreement to conceal should be implied out of the mere fact of conspiracy, and that acts of concealment should be taken as overt acts in furtherance of that implied agreement to *404conceal. Today the Government merely rearranges the argument. It states that the very same acts of concealment should be used as circumstantial evidence from which it can be inferred that there was from the beginning an “actuar’ agreement to conceal. As we see it, the two arguments amount to the same thing: a conspiracy to conceal is being implied from elements which will be present in virtually every conspiracy case, that is, secrecy plus overt acts of concealment.16 There is not a shred of direct evidence in this record to show anything like an express original agreement among the conspirators to continue to act in concert in order to cover up, for their own self-protection, traces of the crime after its commission.
Prior cases in this Court have repeatedly warned that we will view with disfavor attempts to broaden the already pervasive and wide-sweeping nets of conspiracy prosecutions.17 The important considerations of policy behind such warnings need not be again detailed. See Jackson, J., concurring in Krulewitch v. United States, supra. It is these considerations of policy which govern our holding today. As this case was tried, we have before us a typical example of a situation where the Government, *405faced by the bar of the three-year statute, is attempting to open the very floodgates against which Krulewitch warned. We cannot accede to the proposition that the duration of a conspiracy can be indefinitely lengthened merely because the conspiracy is kept a secret, and merely because the conspirators take steps to bury their traces, in order to avoid detection and punishment after the central criminal purpose has been accomplished.
By no means does this mean that acts of concealment can never have significance in furthering a criminal conspiracy. But a vital distinction must be made between acts of concealment done in furtherance of the main criminal objectives of the conspiracy, and acts of concealment done after these central objectives have been attained, for the purpose only of covering up after the crime. Thus the Government argues in its brief that “in the crime of kidnapping, the acts of conspirators in hiding while waiting for ransom would clearly be planned acts of concealment which would be in aid of the conspiracy to kidnap. So here, there can be no doubt that ... all acts of concealment, whether to hide the identity of the conspirators or the action theretofore taken, were unquestionably in furtherance of the initial conspiracy . . . .” We do not think the analogy is valid. Kidnapers in hiding, waiting for ransom, commit acts of concealment in furtherance of the objectives of the conspiracy itself, just as repainting a stolen car would be in furtherance of a conspiracy to steal; in both cases the successful accomplishment of the crime necessitates concealment.18 More closely analogous to our case would be conspiring kidnapers who cover their traces after the main conspiracy is finally ended — i. e., after they have abandoned the kidnaped person and then take care to escape detection. In the latter case, as here, the acts of covering up can by *406themselves indicate nothing more than that the conspirators do not wish to be apprehended — a concomitant, certainly, of every crime since Cain attempted to conceal the murder of Abel from the Lord.
We hold, therefore, that, considering the main objective of the conspiracy to have been the obtaining of “no prosecution” rulings, prosecution was barred by the three-year statute of limitations, since no agreement to conceal the conspiracy after its accomplishment was shown or can be implied on the evidence before us to have been part of the conspiratorial agreement.
II.
In view of how the case was submitted to the jury, we are also unable to accept the Government’s second theory for avoiding the statute of limitations. This theory is (1) that the main objective of the conspiracy was not merely to obtain the initial “no prosecution” rulings in 1948 and 1949, but to obtain final immunity for Gotham and Patullo from criminal tax prosecution; (2) that such immunity was not obtained until 1952, when the statute of limitations had run on the tax-evasion cases which the petitioners conspired to fix;19 (3) that the conspiracy therefore did not end until 1952, when this object was attained; (4) that the acts of concealment within the indictment period were overt acts in furtherance of this conspiracy; and (5) that the prosecution was thus timely.20 In short, the contention is that the agreement *407to conceal was to protect the taxpayers rather than the conspirators, and as such was part of the main conspiracy rather than a subsidiary appendage to it, as under the Government’s first theory.
The Court of Appeals accepted this theory of the case in affirming these convictions. It stated:
“What the fixers had to sell was freedom from criminal prosecution for tax frauds. What the taxpayers bargained for was protection from a tax evasion prosecution.
“This conspiracy is wholly unlike the ordinary illegal scheme in that the jury may well have inferred that the official announcement that there would be no criminal prosecution of the taxpayers *408was merely the delivery of a substantial installment of what appellants agreed to deliver for the huge sums paid. The six-year Statute of Limitations . . . did not run in favor of the taxpayers until some time after the commission of the overt acts relied upon. In the interval there was no assurance, other than continuing efforts by Grünewald, Bolich and the others, that the whole nefarious business might not be brought to light, followed by the revocation of the decision not to criminally prosecute the taxpayers. This is a significant element in the proofs adduced by the government, as concealment of the conspiratorial acts was necessary not only to protect the conspirators from a conspiracy prosecution but also to protect the taxpayers from a tax evasion prosecution.” 233 F. 2d, at 564-565.
We find the legal theory of the Court of Appeals unexceptionable. If the central objective of the conspiracy was to protect the taxpayers from tax-evasion prosecutions, on which the statute of limitations did not run until 1952, and if the 1948 and 1949 “no prosecution” rulings were but an “installment” of what the conspirators aimed to accomplish, then it is clear that the statute of limitations on the conspiracy did not begin to run until 1952, within three years of the indictment.21
Furthermore, we agree with the Court of Appeals that there is evidence in this record which would warrant submission of the case to the jury on the theory that the central object of the conspiracy was not attained in 1948 and 1949, but rather was to immunize the taxpayers completely from prosecution for tax evasion and thus continued into 1952. The many overt acts of concealment occurring after 1949 could easily have been motivated at *409least in part by the purpose of the conspirators to deliver the remaining “installments” owing under the bargain— to wit, the safeguarding of the continued vitality of the “no prosecution” rulings.22 Furthermore, there is evidence showing that from the beginning the aim of the scheme was not restricted to the merely provisional and necessarily precarious “fixing” of the taxpayers’ troubles which was achieved in 1948 and 1949.23 A jury might therefore *410fairly infer that it was part of the conspiratorial agreement that Grünewald and Bolich would make continuing efforts to safeguard the fruits of the partial victories won in 1948 and 1949 by trying to immunize the “no prosecution” rulings from change. In other words, we think a jury could infer from this evidence that the conspirators were prepared and had agreed to engage in further frauds and bribery if necessary in order to maintain in effect the tentative rulings obtained in 1948 and 1949.24
*411If, therefore, the jury could have found that the aim of the conspiratorial agreement was to protect the taxpayers from tax prosecution, and that the overt acts occurring in the indictment period were in furtherance of that aim, we would affirm. We do not think, however, that we may safely assume that the jury so found, for we cannot agree with the Court of Appeals’ holding that this theory of the case was adequately submitted to the jury.
The trial judge’s charge on the problem of the scope and duration of the conspiracy was as follows:
“You will recall that the indictment states, among other things, that it was part of the conspiracy that the defendants and co-conspirators would make ‘continuing efforts to avoid detection and prosecution by any governmental body, executive, legislative, and judicial of tax frauds perpetrated by the defendants and co-conspirators through the use of any means whatsoever including but not limited to . . . the influencing, intimidating, and impeding of prospective witnesses to refrain from disclosing the true facts.’ In other words, the indictment alleges that the conspiracy comprehended within it a conspiracy to conceal the true facts from investigation, should investigation thereafter eventuate. This is an important element of the first count of the indictment which you must take into consideration, inasmuch as the Statute of Limitations on the charge of criminal conspiracy is three years and unless the conspiracy was continuing to a period within three years prior to the date of the indictment, October 25, 1954, and some overt act was performed within that three-year period, the crime, if any, alleged in the first count of the indictment would be outlawed. It is the contention of the government that the conspiracy did not end when the *412taxpayers were advised that there would be no criminal prosecution recommended by the Special Agent’s office, but that an integral part of the entire conspiracy was an agreement to conceal the acts of the conspirators and that when thereafter an investigation was started by Congress and by the Grand Jury in the Eastern District of New York, the conspirators performed overt acts in pursuance of the original conspiracy designed to conceal the true facts; and that these acts occurred within three years prior to the date of the indictment. On this issue, it will be necessary for you to determine whether, beyond a reasonable doubt, you can conclude that the conspiracy was of the nature described in the first count of the indictment and comprehended an agreement to conceal and whether some overt act took place in the period of three years prior to October 25, 1954 to carry out such purpose of the conspiracy.
“To determine whether certain of the alleged overt acts were in furtherance of the object of the conspiracy, you have to determine the duration of the conspiracy. Did it end when the Pattullo [sfc] Modes people and the Gotham Beef people received an assurance of no prosecution from the Bureau of Internal Revenue, or was a part of the conspiracy a continuing agreement to conceal the acts done pursuant thereto? In determining whether a part of the conspiracy was an agreement to continue to conceal the illegal acts after their consummation, you may not imply that such an agreement was part of the conspiracy. You would have to find from the evidence of the acts and declarations of the co-conspirators that there was an understanding or agreement to conceal the conspiracy. If you find that *413such an agreement or understanding to conceal the conspiracy was not a part of the conspiracy to defraud the government, but no more than an afterthought brought to the surface when the co-conspirators were confronted with the Grand Jury and King Committee investigations, then you must find, as a matter of law, that the defendants are not guilty of the crime charged in the first count of the indictment. If you find that the evidence shows, beyond a reasonable doubt, that as a part of a conspiracy to defraud the government, there was an agreement or understanding to conceal the illegal acts and that this too was an objective or part of the conspiracy, then you may find that such understanding was a part of the conspiracy. However, you must additionally determine whether this objective of the conspiracy was known to the defendants. If this objective was known originally by only part of the conspirators but thereafter during the existence of the conspiracy, the scope of the conspiracy was extended so as to include such an agreement to conceal, and if you find that some of the defendants did not know of the expansion to include the agreement to conceal, you may not impute to them the knowledge of their co-conspirators and they could not be found guilty of the crime charged in Count One.”
We are constrained to agree with Judge Frank that this charge did not adequately enlighten the jury as to what they would have to find in order to conclude that the conspiracy was still alive after October 25, 1951. For the charge as given failed completely to distinguish between concealment in order to achieve the central purpose of the conspiracy (that is, the immunization of the taxpayers from tax-evasion prosecution), and concealment intended solely to cover up an already executed crime *414(that is, the obtaining of the “no prosecution” rulings). The jury was never told that these overt acts of concealment could be taken as furthering the conspiracy only if the basic criminal aim of the conspiracy was not yet attained in 1949. On the charge as given, the jury might easily have concluded that the petitioners were guilty even though they found merely (1) that the central aim of the conspiracy was accomplished in 1949, and (2) that the subsequent acts of concealment were motivated exclusively by the conspirators’ fear of a conspiracy prosecution. As far as we know, therefore, the present convictions were based on the impermissible theory discussed in the first part of this opinion — namely, that a subordinate agreement to conceal the conspiracy continued after the central aim of the conspiracy had been accomplished.
Furthermore, if the convictions were based on a finding that the overt acts of concealment were done with the single intention of protecting the conspirators’ own interests, then it is irrelevant that these acts in fact happened to have the effect also of protecting the taxpayers against revocation of the “no prosecution” rulings. For overt acts in a prosecution such as this one are meaningful only if they are within the scope of the conspiratorial agreement. If that agreement did not, expressly or impliedly, contemplate that the conspiracy would continue in its efforts to protect the taxpayers in order to immunize them from tax prosecution, then the scope of the agreement cannot be broadened retroactively by the fact that the conspirators took steps after the conspiracy which incidentally had that effect.
We thus find that the judge’s charge left it open for the jury to convict even though they found that the acts of concealment were motivated purely by the purpose of the conspirators to cover up their already accomplished crime. And this, we think, was fatal error. For the facts in this record are equivocal. The jury might easily have *415concluded that the aim of the conspiracy was accomplished in 1949, and that the overt acts of concealment occurring after that date were done pursuant to the alleged conspiracy to hide the conspirators. As we have said, a conviction on such a theory could not be sustained. Under such circumstances, therefore, it was essential for the judge to charge clearly and unequivocally that on these facts the jury could not infer a continuing conspiracy to conceal the conspiracy, whether actual or implied. Further, it was incumbent on the judge to charge that in order to convict the jury would have to find that the central aim of the conspiracy was to immunize the taxpayers from tax prosecution, that this objective continued in being through 1951, and that the overt acts of concealment proved at trial were at least partly calculated to further this aim.
Since, under the judge’s charge, the convictions on Count 1 might have rested on an impermissible ground, we conclude that they cannot stand, and the petitioners must be given a new trial as to this Count.
III.
What we have held as to the statute of limitations disposes of the conviction of the three petitioners under Count 1, but does not touch Halperin’s conviction on Counts 5, 6, and 7 for violating 18 U. S. C. § 1503.25 As to those Counts, Halperin, who took the stand in his own defense at the trial, contends (a) that the Government was improperly allowed to cross-examine him as to the assertion of his Fifth Amendment privilege before a grand jury investigating this conspiracy, before which he had been called as a witness,26 and (b) that the evidence did *416not justify his conviction on these Counts. For the reasons given hereafter we think that the first contention is well taken, but that the second one is untenable.
In 1952 Halperin was subpoenaed before a Brooklyn grand jury which was investigating corruption in the Bureau of Internal Revenue. Testimony had already been received by the grand jury from the Patullo and Gotham taxpayers, which linked Halperin with the tax-fixing ring. Halperin was asked a series of questions before the grand jury, including, among others, such questions as whether he knew Max Steinberg (an employee of the Bureau of Internal Revenue and a co-defendant in the charge under Count 1); whether he knew Grünewald; whether he had held and delivered escrow money paid to Grünewald by Gotham after the “no prosecution” ruling; and whether he had phoned Grünewald to arrange a meeting between one of his own associates and Bolich. Halperin declined to answer any of these questions, on the ground that the answers would tend to incriminate him and that the Fifth Amendment therefore entitled him not to answer. He repeatedly insisted, before the grand jury that he was wholly innocent, and that he pleaded his Fifth Amendment privilege only on the advice of counsel that answers to these questions might furnish evidence which could be used against him, particularly when he was not represented by counsel and could not cross-examine witnesses before the grand jury.
When the Government cross-examined Halperin at the trial some of the questions which he had been asked before the grand jury were put to him.27 He answered *417each question in a way consistent with innocence. The Government was then allowed, over objection, to bring out in cross-examination that petitioner had pleaded his privilege before the grand jury as to these very questions. Later, in his charge to the jury, the trial judge informed them that petitioner’s Fifth Amendment plea could be taken only as reflecting on his credibility, and that no inference as to guilt or innocence could be drawn therefrom as to Halperin or any co-defendant.28
*418In thus allowing this cross-examination, the District Court relied on Raffel v. United States, 271 U. S. 494, where this Court held that a defendant’s failure to take the stand at his first trial to deny testimony as to an incriminating admission could be used on cross-examination at the second trial, where he did take the stand, to impugn the credibility of his denial of the same admission. In upholding the District Court here, the Court of Appeals likewise relied on Raffel, and also on one of its own earlier decisions.29 Halperin attacks these rulings on these principal grounds: (a) Raffel is distinguishable from the present case; (b) if Raffel permitted this cross-examination, then the trial court erred in refusing to charge, as Halperin requested, that “an innocent man may honestly claim that his answers may tend to incriminate him”; (c) in any case Raffel has impliedly been overruled by Johnson v. United States, 318 U. S. 189; and (d) compelling Halperin to testify before the grand jury, when he had already been marked as a putative defendant, violated his constitutional rights, so that, by analogy to the rule of Weeks v. United States, 232 U. S. 383, his claim of privilege could in no event be used against him. We find that in the circumstances presented here Raffel is not controlling, and that this cross-examination was not permissible.
It is, of course, an elementary rule of evidence that prior statements may be used to impeach the credibility of a criminal defendant or an ordinary witness. But this can be done only if the judge is satisfied that the prior statements are in fact inconsistent. 3 Wigmore, Evi*419dence, § 1040. And so the threshold question here is simply whether, in the circumstances of this case, the trial court erred in holding that Halperin’s plea of the Fifth Amendment privilege before the grand jury involved such inconsistency with any of his trial testimony as to permit its use against him for impeachment purposes.30 We do not think that Raff el is properly to be read either as dispensing with the need for such preliminary scrutiny by the judge, op as establishing as a matter of law that such a prior claim of privilege with reference to a ques*420tion later answered at the trial is always to be deemed to be a prior inconsistent statement, irrespective of the circumstances under which the claim of privilege was made. The issue decided in Raffel came to the Court as a certified question in quite an abstract form,31 and was really centered on the question whether a defendant who takes the stand on a second trial can continue to take advantage of the privilege asserted at the first trial. This Court held, in effect, that when a criminal defendant takes the stand, he waives his privilege completely and becomes subject to cross-examination impeaching his credibility just like any other witness: “His waiver is not partial; having once cast aside the cloak of immunity, he may not resume it at will, whenever cross-examination may be inconvenient or embarrassing.” The Court, in Raffel, did not focus on the question whether the cross-examination there involved was in fact probative in impeaching the defendant’s credibility. In other words, we may assume that under Raffel Halperin in this case was subject to cross-examination impeaching his credibility just like any other witness, and that his Fifth Amendment plea before the grand jury could not carry over any form of immunity when he voluntarily took the stand at the trial. This does not, however, solve the question whether in the particular circumstances of this case the cross-examination should have been excluded because its probative value on the issue of Halperin’s credibility was so negligible as to be far outweighed by its possible impermissible impact on the jury.32 As we consider that in the *421circumstances of the present ease, the trial court, in the exercise of a sound discretion, should have refused to permit this line of cross-examination, we are not faced with the necessity of deciding whether Raffel has been stripped of vitality by the later Johnson case, supra, or of otherwise re-examining Raffel.
We need not tarry long to reiterate our view that, as the two courts below held, no implication of guilt could be drawn from Halperin’s invocation of his Fifth Amendment privilege before the grand jury. Recent re-examination of the history and meaning of the Fifth Amendment has emphasized anew that one of the basic functions of the privilege is to protect innocent men. Griswold, The Fifth Amendment Today, 9-30, 53-82. “Too many, even those who should be better advised, view this privilege as a shelter for wrongdoers. They too readily assume that those who invoke it are either guilty of crime or commit perjury in claiming the privilege.” Ullmann v. United States, 350 U. S. 422, 426. See also Slochower v. Board of Higher Education, 350 U. S. 551, when, at the same Term, this Court said at pp. 557-558: “The privilege serves to protect the innocent who otherwise might be ensnared by ambiguous circumstances.”
When we pass to the issue of credibility, we deem it evident that Halperin’s claim of the Fifth Amendment privilege before the Brooklyn grand jury in response to questions which he answered at the trial was wholly consistent with innocence." Had he answered the questions put to him before the grand jury in the same way he subsequently answered them at trial, this nevertheless *422would have provided the Government with incriminating evidence from his own mouth. For example, had he stated to the grand jury that he knew Grünewald, the admission would have constituted a link between him and a criminal conspiracy, and this would be true even though he was entirely innocent and even though his friendship with Grünewald was above reproach. There was, therefore, as we see it, no inconsistency between Halperin’s statement to the grand jury that answering the question whether he knew Grünewald would tend to furnish incriminating evidence against him, and his subsequent testimony at trial that his acquaintance with Grünewald was free of criminal elements. And the same thing is also true, as we see it, as to his claim of privilege with respect to the other questions asked him before the grand jury and his answers to those same questions when they were put to him at the trial. These conclusions are fortified by a number of other considerations surrounding Hal-perin’s claim of privilege:
First, Halperin repeatedly insisted before the grand jury that he was innocent and that he pleaded his Fifth Amendment privilege solely on the advice of counsel.
Second, the Fifth Amendment claim was made before a grand jury where Halperin was a compelled, and not a voluntary, witness; where he was not represented by counsel; where he could summon no witnesses; and where he had no opportunity to cross-examine witnesses testifying against him. These factors are crucial in weighing whether a plea of the privilege is inconsistent with later exculpatory testimony on the same questions, for the nature of the tribunal which subjects the witness to questioning bears heavily on what inferences can be drawn from a plea of the Fifth Amendment. See Gris-wold, supra, at 62. Innocent men are more likely to plead the privilege in secret proceedings, where they tes*423tify without advice of counsel and without opportunity for cross-examination, than in open court proceedings, where cross-examination and judicially supervised procedure provide safeguards for the establishing of the whole, as against the possibility of merely partial, truth.
Finally, and most important, we cannot deem Halperin’s plea of the Fifth Amendment to be inconsistent with his later testimony at the trial because of the nature of this particular grand-jury proceeding. For, when Hal-perin was questioned before the grand jury, he was quite evidently already considered a potential defendant. The taxpayers whose cases had been “fixed” by the conspiratorial ring had already testified before the grand jury, and they gave there largely the same evidence as they did later, at trial. The scheme was thus in essence already revealed when Halperin was called to testify. Under these circumstances it was evident that Halperin was faced with the possibility of an early indictment, and it was quite natural for him to fear that he was being asked questions for the very purpose of providing evidence against himself. It was thus quite consistent with innocence for him to refuse to provide evidence which could be used by the Government in building its incriminating chain. For many innocent men who know that they are about to be indicted will refuse to help create a case against themselves under circumstances where lack of counsel’s assistance and lack of opportunity for cross-examination will prevent them from bringing out the exculpatory circumstances in the context of which superficially incriminating acts occurred.
We are not unmindful that the question whether a prior statement is sufficiently inconsistent to be allowed to go to the jury on the question of credibility is usually within the discretion of the trial judge. But where such evidentiary matter has grave constitutional overtones, as *424it does here, we feel justified in exercising this Court’s supervisory control to pass on such a question. This is particularly so because in this case the dangers of impermissible use of this evidence far outweighed whatever advantage the Government might have derived from it if properly used. If the jury here followed the judge’s instructions, namely, that the plea of the Fifth Amendment was relevant only to credibility, then the weight to be given this evidence was less than negligible, since, as we have outlined above, there was no true inconsistency involved; it could therefore hardly have affected the Government’s case seriously to exclude the matter completely. On the other hand, the danger that the jury made impermissible use of the testimony by implicitly equating the plea of the Fifth Amendment with guilt is, in light of contemporary history, far from negligible. Weighing these factors, therefore, we feel that we should draw upon our supervisory power over the administration of federal criminal justice in order to rule on the matter. Cf. McNabb v. United States, 318 U. S. 332.
We hold that under the circumstances of this case it was prejudicial error for the trial judge to permit cross-examination of petitioner on his plea of the Fifth Amendment privilege before the grand jury, and that Halperin must therefore be given a new trial on Counts 5, 6, and 7.
Finally, we find no substance to Halperin’s contention that.,he was in effect convicted for advising, as a lawyer, some of the witnesses before the grand jury that they had a right to plead their Fifth Amendment privilege. The evidence against Halperin under these Counts was quite sufficient to make out a case for submission to the jury.
For the reasons given we hold that the judgments below must be reversed, and the cases remanded to the District Court for further proceedings consistent with this opinion.
It is so ordered.
This section provides: “If two or more persons conspire either to commit any offense against the United States, or to defraud the United States, or any agency thereof in any manner or for any purpose, and one or more of such persons do any act to effect the object of the conspiracy, each shall be fined not more than $10,000 or imprisoned not more than five years, or both.”
18 U. S. C. § 1503 provides, in relevant part: “Whoever corruptly . . . endeavors to influence, intimidate, or impede any witness, in any court of the United States or before any United States commissioner or other committing magistrate, or any grand or petit juror, or officer in or of any court of the United States ... in the discharge of his duty ... or corruptly . . . influences, obstructs, or impedes, or endeavors to influence, obstruct, or impede, the due administration of justice, shall be fined not more than $5,000 or imprisoned not more than five years, or both.” Grünewald and Bolich were acquitted on these Counts.
Paragraph 7 of the indictment alleged: “It was a part of the conspiracy that the defendants and co-conspirators would make continuing efforts to avoid detection and prosecution by any governmental body ... of tax frauds perpetrated by the defendants and co-conspirators, through the use of any means whatsoever, including but not limited to, bribery, improper influence and corruption of government employees, the giving of false testimony, [etc.] . . . .”
Paragraph 13 alleged: “It was further a part of the conspiracy that the defendants and co-conspirators at all times would misrepresent, conceal and hide and cause to be misrepresented, concealed and hidden, the acts done pursuant to and the purposes of said conspiracy.”
233 F. 2d, at 559-562.
A “no prosecution” ruling is an internal decision by the investigative branch of the Bureau of Internal Revenue not to press criminal charges against a taxpayer.
The payments were made in cash. In order to raise the money and leave no traces, the taxpayers made unrecorded sales, the profits of which were again unreported income. Further large fees were paid to Halperin and his associates.
Petitioner Bolich was also convicted on Count 2 of the indictment, which charged him and two other Bureau of Internal Revenue employees with conspiracy in violation of 26 U. S. C. § 4047 (e) (4). He was sentenced to three years’ imprisonment and a $5,000 fine on this Count, the prison sentence to run concurrently with the five-year sentence on Count 1. The Court of Appeals held that both Counts related to the same conspiracy, and set aside the separate fine on Count 2.
The governing statute was 18 U. S. C. §3282, which provided: “Except as otherwise expressly provided by law, no person shall be prosecuted, tried, or punished for any offense, not capital, unless the indictment is found . . . within three years next after such offense shall have been committed.”
On September 1, 1954, the statute of limitations was amended to provide for a,five-year limitation period. 68 Stat. 1145, 18 U. S. C. (Supp. Ill) § 3282. Since the amending statute was by its terms made applicable to offenses not barred on its effective date, that is, *397September 1, 1954, it would seem that in fact the crucial date here is September 1, 1951, rather than October 25; in other words, if the conspiracy was still alive after September 1, it was not barred. However, the case was tried on the theory that October 25 was the crucial date, and we so treat it in this opinion. The error, of course, was favorable to the petitioners and was therefore harmless. On the other hand, since we hold that petitioners must have a new trial, the error may be corrected.
See, in general, Lutwak v. United States, 344 U. S. 604; Krulewitch v. United States, 336 U. S. 440; Bollenbach v. United States, 326 U. S. 607; McDonald v. United States, 89 F. 2d 128; United States v. Manton, 107 F. 2d 834; Cousens, Agreement as an Element in Conspiracy, 23 Va. L. Rev. 898; Sayre, Criminal Conspiracy, 35 Harv. L. Rev. 393; Note, 62 Harv. L. Rev. 276; Note, 56 Col. L. Rev. 1216.
In support of this theory, petitioners point to evidence showing that the administrative practice of the Bureau of Internal Revenue was that only recommendations to prosecute would be reviewed at a higher echelon, whereas a determination of no prosecution would, for all practical purposes, end the case. They also emphasize that payment to Grünewald was made under the terms of an escrow which released the money when the “no prosecution” rulings came down.
Petitioners further urge that the acts of concealment occurring after 1949 show at most that a new and separate agreement to conceal was entered into after 1949, an agreement which was not charged in the indictment. Cf. United States v. Siebricht, 59 F. 2d 976. In view of our disposition of the case, we need not deal with this contention.
See n. 3, supra.
336 U. S., at 443-444.
Id., at 455-456.
See 336 U. S., at 444, 455-458; 344 U. S., at 616.
One might cite as an example Grunewald’s attempt at influencing his secretary not to talk to the grand jury, accompanied by an offer to “pay her expenses.” Under the Government’s Krulewitch theory, the argument would have been (in Mr. Justice Jackson’s words) that the “law will impute to the confederates a continuing conspiracy to defeat justice,” and that therefore the other confederates are “bound by another’s unauthorized and unknown . . . bribery of a juror or witness.” But no different result is achieved by saying that the attempted bribe of the witness is evidence from which one can infer an “actual” conspiracy to “defeat justice.” In both cases the essential missing element is a showing that the act was done in furtherance of a prior criminal agreement among the conspirators.
Delli Paoli v. United States, 352 U. S. 232; Lutwak v. United States, supra; Krulewitch v. United States, supra; Bollenbach v. United States, 326 U. S. 607.
See Rettich v. United States, 84 F. 2d 118; McDonald v. United States, 89 F. 2d 128.
The tax evasion cases were governed by a six-year statute of limitations, 26 U. S. C. (1940 ed.) § 3748, which began to run when the last return, pertaining to the year 1946, was filed by the taxpayers.
The Government also suggests a further theory under which this conspiracy could be deemed to have lasted into the indictment period. Under this theory, the central aim of the conspiracy was not specifically to “fix” the tax troubles of Gotham and Patullo, but to engage in the continuing business of fixing any and all tax-fraud *407cases. If this were the aim of the conspiracy, acts of concealment could have been in furtherance of this aim by enabling the ring to stay in business so that it could get new eases. Evidence supporting this theory, says the Government, is that in 1950, after the “no prosecution” rulings in the Patullo and Gotham cases, Halperin engaged in negotiations with another firm which was in tax difficulties. Although these negotiations came to nothing, due to disagreement about the fee to be paid to the conspirators, the incident is presented as evidence that the conspirators were actively soliciting future tax clients in 1950 and were thus still “in business.”
We cannot accept this theory of the Government. The trouble is not only that the theory was never submitted to the jury, but that no overt act done to further the purpose of engaging in “new” business was charged or proved to have occurred after October 25, 1951. If one of the purposes of the conspiracy was to engage in the business of fixing tax cases generally, it must be deemed to have been abandoned in 1951, when investigations of the petitioners started in Congress, since the 1951 and 1952 activities of the conspirators consisted merely of covering up old ventures rather than seeking new ones, and since there is no indication that there was an intent to resume operations after the investigations had ended. Indeed, upon the oral argument the Government seemed to abandon this theory.
The indictment was clearly sufficient to cover submission of this theory to the jury. See n. 3, supra.
One might cite as a typical example an incident in the record occurring in November 1949, 10 months after the “no prosecution” ruling was handed down in the Patullo case. The Special Agent who had been working on the case wrote a final report on it, which stated that Patullo was not prosecuted solely because of Bolich’s decision. This report was sent to Bolich, who thereupon called the Chief of the Conference Section and asked him to write an explanatory memorandum on the case so as to “take a little heat off the situation.” This attempt to “doctor” the report might easily have been motivated not only by fear for himself, but by a purpose to safeguard the “no prosecution” ruling from change in order to maintain the immunity of the taxpayers.
The negotiations between Halperin and his associates and the taxpayers were never very specific as to what exactly was to be accomplished. The tenor of the discussions was that if the taxpayers would hire the mysterious “influential” man in Washington, the matter “would be ended,” the “prosecution end of the case” would be avoided, the matter would be settled “in a civil way without criminal prosecution.” In the same tenor, the accountant of Gotham Beef testified that “nothing at all was to be paid unless the criminal prosecution had been eliminated. It was further understood that they were not at all concerned with the amount of the tax that might result by way of assessment, but it was either that they were completely successful in eliminating criminal prosecution ... or there would be no fee at all.” In other words, there is little indication that it was the specific and narrow end of obtaining the “no prosecution” rulings which was to be the quid pro quo.
This is further buttressed by the fact that the taxpayers were well aware of the precarious nature of the 1948 and 1949 rulings; it is quite clear that they realized that this did not “end” the danger of criminal prosecution. Thus the Patullo taxpayers were aware that the continued investigation of their books for the purposes of civil *410tax liability exposed them to constant danger of “tipping the applecart.” They were warned to “keep their mouths shut,” and a further payment of $25,000 was made for the “boys in New York” so that no one would “raise a fuss about the phony deal that had been put through.” Another Patullo officer testified that, after the “no prosecution” ruling, “we still were not at ease about the thing. We knew that we were elated over the results, but we still were worried about it. There was cooperation to take care of. We had to make this payoff for the New York boys. We were not through with it at that time. We never knew when something else was going to come up. We weren’t through at all. . . . For two years after that we still weren’t through with the thing.” And, referring to the payment for the “New York boys” in 1949: “[W]e never felt too sure about anything because the civil settlement still had to be made and we knew there were people that had to go through it and pass on it and everything, and while this was going on we were told that we would have to get up some more money.”
A jury could thus easily infer that the conspirators’ function did not end in January 1949, and that the conspiratorial agreement contemplated further efforts to immunize the taxpayers from tax prosecution.
It should be mentioned that the Court of Appeals was unanimous in finding that there was sufficient evidence in the record to warrant the submission of the case to the jury on the theory that the central objectives of the conspiracy were not achieved until the statute of limitations ran on the tax-evasion charges. Judge Frank, while dissenting on the ground that the charge to the jury was inadequate in putting the case to the jury on this basis — a view which we share, see infra, p. 413 — agreed that under a proper charge the jury might infer that the conspiracy was still alive through 1951. See 233 F. 2d, at 592-596.
See n. 2, supra.
Grünewald and Bolich also make this contention on their own behalf..
The questions were: (1) Whether petitioner held escrow money which was subsequently delivered to Grünewald; (2) whether petitioner knew Grünewald; (3) whether petitioner made a telephone call to Grünewald relative to an appointment between Bolich and one Davis, a member of the conspiracy; (4) whether petitioner had *417filed a power of attorney in the Glover case; (5) whether he had ever met one Oliphant, an official in the Treasury; (6) whether he knew Steinberg; (7) whether he knew Tobias, the accountant of Gotham Beef; (8) whether he had ever met Grünewald in the Munsey Building in Washington.
The charge as to this point was as follows:
“During the cross examination of one of the defendants, the government questioned the defendant as to his previous statements before the Brooklyn Grand Jury in which he refused to answer certain questions on the ground that answers to them might tend to incriminate him. These questions related to matters similar to those to which the defendant testified at this trial when he took the stand. No witness is required to take the stand or required to give testimony that might tend to incriminate him; but when a defendant takes the stand in his own defense at a trial, it is proper to interrogate him as to previous statements which he may have made under oath concerning the same matter, including his assertion of his constitutional privilege to refuse to testify as to those matters before a grand jury. You may use this evidence of a defendant’s prior assertions of the Fifth Amendment for the sole purpose of ascertaining the weight you choose to give to his present testimony with respect to the same matters upon which he previously invoked his privilege.
“The defendant had the right of asserting the Fifth Amendment when he appeared before the Grand Jury, and I charge you that you are not to draw any inference whatsoever as to the guilt or innocence of the defendant in this case by reason of the fact that he chose to assert his unquestioned right to invoke the Fifth Amendment on that previous occasion. However, it was proper for the Government to question the defendant with respect to his previous invocation of the Fifth Amendment, but you may consider this evidence of *418his prior assertions of the Fifth Amendment only for the purpose of ascertaining the weight you choose to give to his present testimony with respect to the same matters upon which he previously asserted his constitutional privilege. It is not to be considered in a determination of the guilt or innocence of any co-defendant.”
United States v. Gottfried, 165 F. 2d 360, 367.
When the trial court first ruled that the Government could cross-examine as to petitioner’s Fifth Amendment plea, it did not do so on the grounds of inconsistency reflecting on credibility. In fact the implication to be drawn from the record is that the court at that time felt that the jury might use this evidence for any purpose at all, including the drawing of inferences as to guilt or innocence. When the Government first embarked on this method of cross-examination, the judge overruled objections in these words:
“The Court: I know the Government’s position. As I see it, Mr. Corbin [a defense attorney], no witness can be compelled to testify against himself. The witness is called before the grand jury and the answer was, I refuse to answer something on the ground that if I answer that question it will incriminate me.
“Mr. Corbin: Tend to incriminate.
“The Court: Or tend to incriminate. A witness can make that statement. No witness has to take the witness stand, as I understand the law and if a witness has so stated, then he could not be compelled to take the stand here, but if a witness voluntarily takes the stand and is asked in a previous proceeding did you say any testimony on this subject would incriminate you, that can be considered by the jury for such benefit or such worth as the jury may want to give it.”
When the defendants asked that at the very least the use of this evidence be restricted to the question of credibility, the judge contented himself with asking for a memorandum of law on the subject.. Thus, although later, in the charge to the jury, the matter was specifically restricted to the issue of credibility, there was no inquiry by the judge at the time of the initial admission of this evidence as to whether a sufficient showing of inconsistency had been made.
The certified question was: “Was it error to require the defendant, Raffel, offering himself as a witness upon the second trial, to disclose that he had not testified as a witness in his own behalf upon the first trial?” 271 U. S., at 496.
In Raffel this Court assumed that the defendant’s failure to testify at the first trial could not be used as evidence of guilt in the second trial, 271 U. S., at 497. The Court further stated that “the *421trial judge might appropriately instruct the jury that the failure of the defendant to take the stand in his own behalf is not in itself to be taken as an admission of the truth of the testimony which he did not deny.” As already indicated, p. 418; supra, here the trial judge refused to charge that “an innocent man may honestly claim that his answers may tend to incriminate him.”