*32OPINION OF THE COURT
Gabrielli, J.This case requires us to determine whether the testimony of an admitted accomplice may artificially be divided into testimony concerning events which occurred before the witness became an accomplice and testimony about incidents which took place after the witness became an accomplice, so that the prior testimony can serve as corroboration for the latter testimony. The accomplice corroboration rule cannot be so readily evaded, for an accomplice simply cannot corroborate his own testimony. Any other result would serve only to eviscerate the accomplice corroboration requirement imposed by CPL 60.22 (subd 1).
The prosecutions involved in this appeal are based on a scheme devised by several New York City police officers to obtain funds from gamblers in return for a guarantee that the operations of those gamblers would not be curtailed by police interference. Following an extensive jury trial, each defendant involved in this appeal was convicted of one count of bribe receiving, one count of receiving a reward for official misconduct, two counts of conspiracy, and one count of official misconduct. On an appeal by the convicted defendants, the Appellate Division affirmed the convictions of defendants Mel-nick, White, Bergold and Fortuna. The convictions of defendants Reitano, Maroney, Brown, Conti, Carter, Cona, Auletta, Zummo, Mattina, Callis and Greene, however, were reversed by the Appellate Division on the ground that the evidence of those defendants’ guilt was comprised solely of uncorroborated accomplice testimony. For the reasons discussed below, we conclude that the order of the Appellate Division should be modified by reinstating the convictions of defendants Reitano, Maroney, Brown, Conti, Carter, Cona and Auletta, and otherwise affirmed.
The convictions of defendants Reitano, Maroney, Brown, Conti, Carter, Cona and Auletta were based on the uncorroborated testimony of Police Officer Buchalski, originally a conceded member of the conspiracy who eventually became a police agent.1 The Appellate Division concluded that these *33convictions could not stand due to the absence of corroboration other than Buchalski’s own testimony. Due to our limited powers of review we may not reach this issue. To create and preserve a question of law amenable to appellate review, a defendant in a criminal case normally must raise that issue before the court of original jurisdiction (CPL 470.05, subd 2; compare People v Michael, 48 NY2d 1, with People v Robinson, 36 NY2d 224). In the instant case, no objection was made to that portion of the court’s charge which dealt with the application of the accomplice corroboration rule to Buchalski’s testimony concerning these defendants; hence, under settled rules of law and statutory construction, these defendants failed to preserve a question of law as to the correctness of that portion of the charge.2
Unlike the Court of Appeals, the Appellate Division as an intermediate court of review has broad power to review questions of fact and discretion as well as questions of law (see CPL 470.15, 470.30). Hence, the Appellate Division, if it deems it appropriate, may exercise its discretionary power to review alleged errors even in the absence of that timely objection which is necessary to create a question of law. Where the Appellate Division exercises this discretion, however, and reverses a conviction on the basis of an issue not preserved, its order is then grounded at least in part upon the exercise of that discretion and is thus not appealable to this court (see People v Johnson, 47 NY2d 124; People v Williams, 31 NY2d 151).
In the instant case, the Appellate Division reviewed defendants’ claim of error with respect to the accomplice corroboration charge given concerning Buchalski’s testimony, *34despite the defendants’ failure to timely object to that portion of the charge. Had the Appellate Division’s decision to review this claim been based on an exercise of discretion, we would have been compelled to dismiss the appeal insofar as it pertains to the Appellate Division reversal of the convictions of these defendants. As it is, however, the Appellate Division erroneously concluded as a matter of law that a question of law was preserved with respect to the corroboration of Buchalski’s testimony despite defendants’ failure to object. Hence, the Appellate Division order was actually based on a determination of law, albeit an erroneous one, and is thus not subject to dismissal. Our power to review, however, is limited to that question of law, and after concluding that the Appellate Division erred in determining that a question of law was presented, our review must stop at that point and we must remit the matter to the Appellate Division for that court to determine whether it would be appropriate to exercise its broader review power.
A different situation exists, however, with respect to defendants Zummo and Mattina. Their convictions were based upon the uncorroborated testimony of Police Officer O’Brien. O’Brien’s initial contacts with this criminal enterprise arose in the course of his activities as an undercover police agent assigned to investigate possible police corruption. Had he remained faithful to his duty, there would of course be no need to corroborate his testimony. This is so because an undercover police agent is simply not an accomplice, due to the absence of any criminal intent. As it is, however, O’Brien eventually succumbed to temptation, eschewed his obligations as an undercover agent, and became an active and actual member of the conspiracy. Although the trial court charged the jury that O’Brien was an accomplice as a matter of law after this conversion and that his testimony concerning incidents which occurred while he was an accomplice must be corroborated, the court refused to charge that O’Brien’s testimony about prior events was also subject to the corroboration requirement. By timely requesting such a charge, these defendants created a question of law amenable to appellate review; hence the determination of the Appellate Division as to these defendants was made on the law alone and is subject to our review. For the reasons discussed below, we conclude that the Appellate Division correctly set aside the convictions of defendants Zummo and Mattina.
*35An accomplice is "a witness in a criminal action who, according to evidence adduced in such action, may reasonably be considered to have participated in: (a) The offense charged; or (b) An offense based upon the same or some of the same facts or conduct which constitute the offense charged” (CPL 60.22, subd 2). The prosecutions underlying this appeal were premised upon the defendants’ participation in an ongoing criminal enterprise pursuant to which police officers solicited bribes from gamblers on a regular basis in return for "protection” from police interference with the gamblers’ operations. It is conceded that O’Brien fully participated in this scheme for a considerable period of time following his abandonment of his duties as an undercover agent. Hence, at the very least, he necessarily participated in "[a]n offense based upon the same or some of the same facts or conduct which constitute the offense charged”. Having reached this conclusion, we must hold that, in accord with the plain language of the law, the defendants could not be convicted on the basis of O’Brien’s uncorroborated testimony. There exists no justification for distinguishing between testimony pertaining to different periods of time as a basis for application of the accomplice corroboration rule. The effect of a contrary decision would be to make it possible to avoid the accomplice corroboration rule in any situation in which an admitted accomplice did not participate in all of the crimes for which the defendants are tried. In such a case, the testimony of the accomplice concerning incidents which he observed but in which he did not participate, under the approach urged by the People, would not be subject to the corroboration requirement and could indeed serve as corroboration for the witness’ other testimony. So viewed, the result in such a case would be that which the accomplice rule was specifically designed to avoid: "conviction [based] solely upon the testimony of persons who are in some way criminally implicated in * * * the general conduct or factual transaction on trial” (Denzer, Practice Commentary, McKinney’s Cons Laws of NY, Book 11A, CPL 60.22, pp 194-195).
The accomplice corroboration rule is premised upon a legislative determination that the testimony, of individuals who may themselves be criminally liable is inherently suspect. This is deemed to be true because such individuals may be subject to pressures impelling them to color testimony in order to protect themselves by belittling the actual extent of *36their involvement in the criminal enterprise at the expense of others. In a similar vein, a person who agrees to turn State’s evidence may believe it to be in his best interests to embellish the truth when testifying about the defendants in order to secure the approval of officials who have the authority to prosecute the accomplice should they not be satisfied with his cooperation. In short, the accomplice corroboration rule entails a legislative recognition that, as a result of these various pressures, both real and imagined, to which an informant may be subject, such testimony is somewhat suspect. At the same time, such testimony is often a necessary predicate for successful criminal prosecutions. Whatever the wisdom of the solution the Legislature has developed in order to resolve this dilemma, the rule is plain and must be properly applied by the courts.
The dissenting opinion is impliedly premised on the perception that a person who is subject to criminal liability for some but not all of the crimes committed in the course of an ongoing criminal enterprise will be less inclined to color the truth in testifying about those incidents for which he cannot be liable. This view is unrealistic in the extreme, for the desire of the witness to diminish his own potential criminal liability and to please the prosecutor will inevitably affect his entire testimony, since it is the result of the entire prosecution which will be of concern to the informant. More importantly, even if the dissenters’ premise were true, it would be irrelevant. The basis for the accomplice corroboration rule, quite simply, is mistrust of an informant’s testimony about matters relevant to his potential criminal liability. Thus, the rule requires corroboration from some source other than the informant himself, in order to support a conviction based upon the informant’s testimony. The fact that part of an informant’s testimony may not be concerned with matters for which he himself is criminally liable in no way militates against the suspect nature of his testimony taken as a whole. The type of bootstrapping which the dissent would endorse in this case is in clear conflict with the purpose and intent of the accomplice corroboration rule, since it would result in convictions based solely upon the testimony of an accomplice. Any change in this rule should come from the Legislature, not the courts.
In conclusion, we note that our decision today warrants none of the spectres of "shackling the forces of law enforce*37ment” or "depriving] law enforcement of a vital and necessary investigative technique” envisaged by the dissenters. The "use of the turncoat conspirator” will continue to be fully available. We require only that the evidence thereby produced be supported by corroborative evidence tending to connect the defendant with the commission of the crime. It should be emphasized, moreover, that the accomplice corroboration rule obviously is not applicable to the testimony of all or even most undercover police agents; corroboration is necessary only with respect to the testimony of those prosecution witnesses who actually were accomplices. This is not a crippling requirement and serves to assure the reliability of evidence furnished by. a witness who must be perceived as possibly laboring under considerable inducements to favor the prosecution.
Turning to the remaining issues raised on this appeal, we find that the convictions of defendants Melnick, White, Bergold and Fortuna were properly affirmed by the Appellate Division, for the reasons stated in the majority opinion at the Appellate Division. As to defendants Callis and Greene, we agree with the Appellate Division that there is insufficient evidence in the record to corroborate the accomplice testimony implicating those defendants.3 We find the other contentions raised by defendants to be without merit.
Accordingly, the order of the Appellate Division should be modified by reinstating the convictions of defendants Reitano, Maroney, Brown, Conti, Carter, Cona and Auletta and the case as to those defendants should be remitted to the Appellate Division for that court to determine whether to exercise its discretionary power to review defendants’ accomplice corroboration claim in the absence of any timely objection, and, as so modified, should be affirmed.
. Tape recordings of various conversations allegedly entered into between Buchalski and these defendants were also admitted into evidence. Since the only evidence showing that the voices on the tapes were those of these defendants, however, was the uncorroborated testimony of Buchalski, the tapes may not serve as corroboration of Buchalski’s other testimony. This is to be contrasted to the situation of defendants *33White and Melnick, who admitted that the voices on the tapes were theirs. By so. stating, they themselves supplied the necessary corroboration.
. It has been suggested that this issue was preserved for appellate review because of a motion to dismiss for insufficient evidence. Our examination of the record, however, persuades us that this motion, although formally made on behalf of all defendants, was not specifically directed at the failure to require corroboration for Buchalski’s testimony about these defendants. Conspicuous by its absence is any suggestion that the People’s proof as to these seven defendants was insufficient due to the failure to corroborate Buchalski’s testimony. Indeed, in furtherance of a legal argument not relevant to this appeal, the defendants argued strenuously that Buchalski should not be considered an accomplice after he agreed to serve as a police agent. At any rate, the failure to specify the lack of corroboration for Buchalski’s testimony as a basis for the motion to dismiss the case against these seven defendants precludes later use of that motion as a vehicle for creating a question of law on this point (see People v Spiegel, 48 NY2d 647, 648).
. We note that a timely motion was made to dismiss the charges against these defendants due to the lack of corroboration for the accomplice testimony upon which the case against them was based.