In People v Rodriguez (38 NY2d 95), the Court of Appeals reaffirmed the principle that where a defendant comes forward with evidence, "his failure to call an available witness who is under defendant’s control and has information material to the case may be brought to the jurors’ attention for their consideration” (People v Rodriguez, supra, p 98). The court in Rodriguez did not squarely address the question primarily raised on this appeal—the nature and form of the instruction appropriately to be given to the jury under the applicable circumstances.
In this case the trial court charged the jury with regard to *632the defendant’s failure to call a certain witness to testify on the justification defense that "the strongest inference may be drawn against the defendant which opposing evidence in the record permits.” In my opinion, this instruction permitted far too strong an inference to be drawn against the defendant, is fundamentally wrong in principle, and, if accepted as an appropriate instruction, would have serious and harmful consequences in the trial of many criminal cases.
Analysis of the issue appropriately starts with the observation from Richardson on Evidence (10th ed, § 92, pp 67-68), quoted in Judge Silverman’s opinion for the court: "While the authorities agree that, in a proper case, the failure of a party to call a witness will permit an unfavorable inference, there is no such accord in New York as to the precise nature of the unfavorable inference which may be drawn.”
Two formulas, each with variations, have been developed in the New York cases. One in substance provides that a jury may infer that had the witness been called his testimony would have been unfavorable. The other favors an instruction to the effect that the jury may construe the evidence already in the case most strongly against the party who might have called the witness to contradict or explain that evidence. The latter phrasing has apparently become the more accepted one in New York (see Noce v Kaufman, 2 NY2d 347, 353; see, also, PJI 1:75), although the "unfavorable inference” version appears to be more widely accepted in other jurisdictions. (See 2 Wigmore, Evidence [3d ed], § 285; McCormick, Evidence [2d ed, 1972], § 272.)
It seems clear that these alternative approaches, with their accompanying variations, have been developed primarily in civil cases. Even with regard to such cases the established charges have come under increasing criticism. Thus, in Fisch on New York Evidence (2d ed, § 1126), the following comment is made with regard to the "strongest possible inference” instruction (p 638): "Strongly condemned as unrealistic and self-defeating, the value of this rule as an intelligible guide has been challenged on the ground that the jury cannot follow and apply the complexities of a charge in which it is embodied. Moreover, it is felt that since the situation requires the use of common experience rather than a mechanical rule the inference created by a failure to call a witness is not capable of being forced into any prescribed formula. It has thus been suggested that counsel be permitted to comment on the failure *633to call a witness and that the jury be allowed to draw whatever inference, if any, is deemed warranted.”
In his authoritative treatise, McCormick said (§ 272, p 657): "Despite the plenitude of cases supporting the inference, caution in allowing it is suggested with increasing frequency. This counsel of caution is reinforced by several factors. Possible conjecture or ambiguity of inference is often present. The possibility that the inference may be drawn invites waste of time in calling unnecessary witnesses or in presenting evidence to explain why they were not called. Anticipating that the inference may be invoked entails substantial possibilities of surprise.”
These observations by leading authorities are of more than academic interest. It is my personal observation that, for the reasons suggested, many experienced Trial Judges in civil cases often find the established instructions far too strong and use other and much less explicit and directive terminology.
Whatever may be the merits of either of the established charges with regard to civil cases, their use against defendants in criminal cases raises disturbing questions. The inescapable result in many criminal cases will surely be to erode effectively the prosecutor’s traditional burden to prove guilt beyond a reasonable doubt and to dissuade defendants from testifying or introducing evidence on their own behalf.
In the usual criminal case, as in this one, the defense witness whose failure to testify gives rise to an adverse inference will most often be a relative or a close personal friend. Ironically, the very circumstance that provides a basis for the instruction often reduces significantly and sometimes dramatically the probative value of testimony by that witness.
It is surely a familar observation that juries are not quick to believe testimony offered on behalf of a defendant in a criminal case by a parent, spouse or, as here a girlfriend. Where, as is often the case, the witness is believed by defense counsel to be vulnerable to cross-examination, sometimes on collateral matters, a lawyer may reasonably conclude that a witness on balance would not be helpful even though the witness would in fact give testimony favorable to the defendant.
If the instruction given here is approved, defense counsel will be confronted in many cases with several exceedingly distasteful alternatives: (1) to call a witness whose relationship to the defendant will invite skepticism and whose testimony on balance may be harmful; (2) not to call such a *634witness and face the impact of a crushing adverse inference charge by the court; and (3) not to offer any evidence whatever on behalf of the defendant.
I do not believe that the interests of justice will be served by confronting lawyers with these choices.
Nor is the instant case free from the "ambiguity of inference” alluded to by McCormick. It is true that the witness was the defendant’s girlfriend at the time the events resulting in the criminal charge occurred. However, the record discloses that he had been in prison for well over a year at the time of the trial. While it appears that he communicated with her by telephone and letter during that period, and may have received visits from her, it is by no means apparent that the relationship at the time of trial was free from uncertainty.
Nor do we learn from the record whether the police and the District Attorney, alerted to her alleged role by the defendant when he was arrested, ever sought to question her and, if so, whether she had answered the questions, and what answers, if any, she had given. In short, here, as in many cases, although the circumstances permit the inference that she was subject to the defendant’s control, it is not at all clear that this was indeed the fact.
The principal issue presented on the trial was one of justification. Although there were circumstances that supported the complaining witness’ version of the events, as note in Justice Silverman’s opinion, the issue was not free from doubt. The testimony of the complaining witness wholly failed to explain why the defendant suddenly started to shoot him, an omission that may well have raised doubts in the minds of jurors as to his credibility on that point and perhaps on others as well.
It was, of course, the People’s duty to establish the defendant’s guilt beyond a reasonable doubt. As the factual issue was presented to the jury, the effect of the court’s charge that the jury may draw the strongest inference against the defendant permitted by opposing evidence (that is, the testimony of the complaining witness) effectively obliterated the prosecutor’s burden of proof, and would inevitably do so in any criminal trial in which the missing witness is thought to have testimony pertinent to the central issue.
It seems to me inescapable that the charge used, if it becomes the accepted law, will often result in defendants not *635testifying or offering pertinent evidence on their behalf in order to avoid its impact.
The appropriate charge to be used with regard to a defendant in a criminal case seems to me the very one that was used by the trial court in People v Rodriguez (38 NY2d 95, 101-102, supra), described in the court’s opinion as follows: "The supplementary charge, brief as it was, said little more than that the failure to call the wife was a matter the jury could 'consider’. It was correct as far as it went. At worst, it was innocuous.”
It is true that in the course of its opinion in People v Rodriguez (supra), the Court of Appeals cited Noce v Kaufman (supra), and other civil cases in which the "strongest possible inference” charge was used. However, in the Rodriguez opinion itself the principle was consistently phrased in terms of the jury’s right to consider.
That, in my opinion, is the most that should be said in a criminal case with regard to a defendant. And for reasons set forth above, I believe it a grave error to permit the kind of sweeping inference here charged to the jury.
The defendant’s conviction of assault in the first degree should be reversed and the case remanded for a new trial.
All concur except Kupferman, J. P., and Markewich, J., who dissent in part and would affirm the judgment in toto, and Birns and Sandler, JJ., who dissent in part in an opinion by Sandler, J., and would reverse the conviction and remand the case for a new trial, but concur in reduction of sentence.
Judgment, Supreme Court, Bronx County, rendered on January 31, 1978, modified, as a matter of discretion in the interest of justice, and the sentence of the defendant is reduced to an indeterminate term of imprisonment of not less than 3 nor more than 15 years, and the judgment is otherwise affirmed.