OPINION OF THE COURT
Duncan, Chief Judge:Tried by a general court-martial convened at Fort Leavenworth, Kansas, the accused was convicted of conspiracy to commit murder and premeditated murder. He was sentenced to life imprisonment. Intermediate appellate action has resulted in approval only of findings of guilty of unpremeditated murder and a sentence to confinement at hard labor for 45 years. At this level, the accused contends that he was prejudiced by the military judge’s erroneous instructions regarding the effect of evidence of other misconduct, his failure to instruct on the credibility to be accorded an accomplice’s testimony, and the competency of that accomplice as a witness in light of the terms of a pretrial agreement with the convening authority that allegedly caused the accomplice’s testimony to be shaped in a certain direction.
I
The tragic occurrence involved in the charge now before us had its genesis in an altercation between the accused and his victim on the basketball court at the Local Parolee Unit, a part of the United States Disciplinary Barracks. The two men were quickly parted, but the fight was renewed in a barracks latrine. There the accused suffered facial injuries at the hands of his victim. The Government’s evidence indicates that the accused, Prisoner Burns (his closest friend) and Prisoner Fuqua determined to seek vengeance. In order to avoid detection, it was decided to wait until after the prisoner count had been completed and lights turned out in the bay in which the victim slept. Fuqua was then to act as a lookout, while Burns and the accused, one armed with an iron pipe and the other with an iron bar, entered the victim’s bay and assaulted him.
This plan was substantially executed, and the victim died from the effects of a subdural hematoma, having had his skull fractured in five places during the attack on him. Several eyewitnesses identified both the accused and Burns as having carried out the assault, testifying vividly to the manner in which the accused raised his weapon above his shoulders and struck the victim’s head with swift chopping blows. One witness heard the accused state, as he departed, " T told you I’d get you.’ ”
In accordance with his pretrial agreement with the convening authority, Fu-qua agreed to testify against the accused. While identifying the weapons and establishing the manner in which the victim’s beating was planned, he minimized the accused’s role in the preparations, attributing the leadership of the group to Burns. He described the accused as a passive participant in the planning and stated that their objective was only to "tighten up” the victim. He declared that there was never any intention to kill the victim.
In his final argument, defense counsel adopted Fuqua’s version of the events which transpired prior to the assault and urged that the court accept Fuqua’s testimony as to the appellant’s role in the slaying. Bearing this state of the record in mind, we turn to the issues presented.
II
A defense witness testified that the accused’s reputation for peaceableness was good. On cross-examination, the trial counsel inquired whether he was aware that the accused had previously been convicted by general court-martial for various offenses and whether such knowledge would affect his opinion as to the accused’s reputation. At the close of *6the trial, the military judge advised the court that the evidence of the accused’s previous convictions could be considered on the issues of premeditation and intent to kill or inflict great bodily harm involved in the offenses of premeditated and unpremeditated murder.
The Government concedes that the instruction was erroneous and that consideration of the previous convictions should have been limited to their tendency to impeach the witness’ testimony. See United States v Dixon, 17 USCMA 423, 38 CMR 221 (1968); United States v Back, 13 USCMA 568, 33 CMR 100 (1963); United States v Bryant, 12 USCMA 111, 30 CMR 111 (1961). It argues nonetheless that the effect of the error is to be measured by the gauge of specific prejudice and that, in this case, the accused was not harmed. s
We agree with the Government’s contention. The lack of a proper instruction on the effect of uncharged misconduct is not generally prejudicial. Its impact must be measured according to whether there is a fair risk that the accused was harmed by its improper delineation. United States v Johnson, 17 USCMA 479, 38 CMR 277 (1968); United States v Kirby, 16 USCMA 517, 37 CMR 137 (1967); United States v Satey, 16 USCMA 100, 36 CMR 256 (1966).
As was stated in United States v Back, supra:
On the one hand, evidence of accused’s guilt may be such that the failure to restrict proof of other misconduct may be fairly said to have weighed not at all in connection with the findings and sentence. On the other, the record may present a fair risk that the fact finders accorded weight on the merits to the matter. We cannot lay down any precise measure for answering this subsidiary question, which, of course, depends so much upon the circumstances of the individual case.
13 USCMA at 571, 33 CMR at 103.
In the present case, the evidence of the accused’s guilt of unpremeditated murder is overwhelming. It was uncontested that he was seeking revenge. Un-contradicted testimony of several eyewitnesses identified him as standing over his sleeping victim and striking him several times on the head with a pipe-like object. Their graphic description of the manner in which the blows were struck leaves no doubt that either death or grievous bodily harm was intended. There is likewise no question but that the victim died of the wounds thus inflicted on him. On the other hand, no circumstances were presented that would tend to permit the factfinders to mitigate the degree of homicide to manslaughter or that tended to establish the existence of any affirmative defense.
In light of these considerations, we necessarily conclude "the evidence is . . . such as to impel a morally certain conviction of accused’s guilt in the minds of the fact finders,” United States v Back, supra at 572, 33 CMR at 194, and that there is no fair risk the evidence of his previous convictions played any part in their deliberations. We conclude that the erroneous instruction was not prejudicial.
Ill
Accused also contends that it was prejudicial error for the military judge to fail to instruct sua sponte on the credibility to be accorded accomplice testimony. He contends that Fuqua, as an accomplice, was the Government’s single most important witness, that Fuqua testified to the planning of the deadly assault, identified the murder weapons, and provided the proof establishing the necessary intent to kill or inflict great bodily harm. Thus, he contends that this case is the exceptional contest in which an accomplice serves as a crucial Government witness and requires an instruction on the effect of his testimony without request from the defense counsel. See United States v Lell, 16 USCMA 161, 36 CMR 317 (1966); United States v Stephen, 15 USCMA 314, 35 CMR 286 (1965).
We do not question the accuracy of the accused’s analysis of the precedents of this Court regarding the duty of a military judge to instruct on accomplice testimony. Both Lell and Stephen make clear that the judge must instruct sua sponte on the effect of accomplice testimony when such testimony is of pivotal importance to the Government’s case. *7These authorities also note that such situations are exceptional, that is to say, that the failure of counsel to request such an instruction is fatal to his argument on appeal unless the circumstances are such that this Court must act to prevent a manifest miscarriage of justice. Typically, the situation presented is one in which the accomplice is the crucial prosecution witness on whose credibility the outcome of the case hinges. United States v Lell, supra; United States v Stephen, supra at 317, 35 CMR at 290. This is not such a case.
In the present record, Fuqua’s testimony as to the planning of the incident tended to establish little more than that the parties intended to administer a beating to the victim. He characterized the accused as a passive participant in the events and specifically denied that the parties intended to kill. In contrast to his testimony, other witnesses identified the accused as participating in the actual assault and vividly described his actions in a manner that left little doubt as to his intent. It was these witnesses who were crucial to the prosecution obtaining a conviction of murder rather than Fuqua, and none of them were accomplices.
We also note that trial defense counsel was apparently of the same view for, in his final argument, he urged the court members to adopt the view of the events supported by Fuqua’s testimony as the only credible detailing of the crime. In light of this, we áre not at all surprised that he failed to request an instruction that would have undercut Fuqua’s credibility and left the court only with the much stronger case offered by the eyewitness testimony.
In light of these considerations, we are convinced that this is not one of the exceptional cases in which the military judge is required to instruct sua sponte on accomplice testimony.
IV
Finally, the defense contends that it was a violation of public policy to permit Fuqua to testify in light of his pretrial agreement with the convening authority to testify against the accused in a certain, predetermined manner. The accused argues that the agreement made Fuqua an incompetent witness. The Court of Military Review agreed with the defense contention and, as a result, set aside his conviction of conspiracy. At this level, counsel argue that this action was insufficient to cure the error and that the approved findings of guilty of unpremeditated murder must also fall.
The agreement in question was one in which the witness agreed to plead guilty to a charge of conspiracy to commit aggravated assault in return for dismissal of certain other charges. In addition, Fuqua stated in his offer of the agreement, found in Appellate Exhibit I of his record of trial:
I also offer to render testimony in the cases of US v Burns and US v Gilliam which would establish conspiracy and premeditation by such individuals and would be able to identify the implements used by Private E-l William P. Burns and Sentenced Prisoner Donovan B. Gilliam.
United States v Fuqua, CM 427573 (ACMR Dec. 21, 1972).
It is this provision of the agreement with which appellant quarrels. The defense properly concedes that the mere existence of a grant of immunity or pretrial agreement does not render a witness incompetent. United States v Moffett, 10 USCMA 169, 27 CMR 243 (1959); United States v White, 10 USCMA 63, 27 CMR 137 (1958). But it also points out that the testimony of a witness at trial may not be conditioned by the terms of a grant of immunity or a pretrial agreement. Relying on our decisions in United States v Conway, 20 USCMA 99, 42 CMR 291 (1970); United States v Stoltz, 14 USCMA 461, 34 CMR 241 (1963); and United States v Scoles, 14 USCMA 14, 33 CMR 226 (1963), it argues that Fuqua’s agreement falls within the prohibited area.
In United States v Scoles, the witness’ agreement with the convening authority included a provision for the reduction of his sentence by 1 year for each occasion on which he testified against his co-accused. We concluded that this contingency agreement was contrary to public policy as offering "an almost irresistible temptation to a confessedly guilty party *8to testify falsely in order to escape the adjudged consequences of his own misconduct.” 14 USCMA at 20, 33 CMR at 232.
In United States v Stoltz, the witness was granted immunity from prosecution, premised upon his testifying in accordance with a pretrial statement that was set forth in the grant of immunity. His testimony was required to include the matters contained in that statement. We reversed, for the grant required the witness "to testify under oath to the particular matters extracted from his written pretrial statement . . . regardless of the truth of the matters concerning which he had knowledge.” 14 USCMA at 464, 34 CMR at 244. Only in this way could he maintain his immunity.
In United States v Conway, the witness was required by his agreement to testify against the accused and " 'follow the statement that I previously made, and if I don’t, I’ll receive a general court-martial.’ ” 20 USCMA at 100, 42 CMR at 292. Relying on Stoltz and Scoles, we reversed, holding that; absent evidence of a complete understanding between the Government and the witness that he was to testify only truthfully, the agreement made him an incompetent witness.
The agreement in the present case resembles those in the cited authorities. While it does not go so far as the conditions imposed on the witnesses in those cases, it still required Fuqua to testify in a particular manner, i.e., to identify the murder weapons and to "establish” conspiracy and premeditation. In return for dismissal of certain of the charges against him, Fuqua bound himself to "establish” these matters on behalf of the Government, without regard for the sanctity of his oath as a witness. Such limitations and conditions on the giving of testimony should play no part in a pretrial agreement. It was error to include them, and the Court of Military Review properly concluded that dismissal of the conspiracy charge was necessary.
It is another matter, however, to say that the error was also fatal to the approved findings of guilty of unpremeditated murder. Adverting once more to the clear and convincing evidence of the accused’s guilt without regard to Fuqua’s declarations and the defense adoption at trial of his version of the events as the one most favorable to the accused, we see no prejudice from the presentation of Fuqua’s testimony insofar as the offense of unpremeditated murder is concerned. We nevertheless reiterate our belief that pretrial agreements should not be concerned with the scope of a prospective witness’ testimony, but should be limited to the fact that he has agreed to testify in a particular case.
The decision of the U. S. Army Court of Military Review is affirmed.
Judge Quinn and Senior Judge Ferguson concur.