Opinion of the Court
EVERETT, Chief Judge:
Notwithstanding Sutton’s pleas of not guilty, a general court-martial convicted him of wrongful use of cocaine in August 1987, in violation of Article 112a, Uniform Code of Military Justice, 10 USC § 912a. It then sentenced him to a bad-conduct discharge, confinement and forfeiture of $200.00 pay per month for 6 months, and reduction to the lowest enlisted grade. The convening authority approved these results, and the Court of Military Review affirmed by a divided vote. 27 MJ 578 (1988).
In this Court, appellant challenges the correctness of the military judge’s pretrial ruling that a “negative” urinalysis report could be used by the prosecution in cross-examining potential defense character witnesses and in cross-examining appellant, if he subsequently took the stand and denied ever having used drugs. Moreover, appellant attacks the majority’s holding in the Court of Military Review that he waived appellate review of the judge’s ruling by his failure to testify in this manner or to offer any defense character witnesses. 28 MJ 458 (1989). On both contentions, we agree with appellant.
I
In a pretrial session held under Article 39(a), UCMJ, 10 USC § 839(a), appellant moved to suppress “any reference” to a urinalysis performed on a specimen that appellant had produced by command direction on September 25, 1987. The laboratory report in question and other evidence indicated that appellant’s specimen had been tested by radioimmunoassay (RIA); that this procedure had indicated a presence of cocaine below the 300 ng/ml minimum required by Department of Defense directives for a “positive” report; and that, consequently, the specimen had been discarded without confirmation by the more sophisticated and accurate gas chromatograph-mass spectrometry (GC/MS) procedure.
Essentially, the defense argued that, because the screening urinalysis had resulted in a “negative” report, without any subsequent confirmation testing, this report should not be used by the Government during cross-examination or for rebuttal. In response, the prosecution urged that, depending on how the defense case unfolded, it could “foresee a number of uses that could be made of this evidence” — such as [13]*13to impeach or to show Sutton’s knowledge or lack of mistake.1
After hearing substantial evidence concerning reliability of the RIA procedure, the military judge found that the test was of the type upon which experts could reasonably rely and that, accordingly, an expert could use the test as a basis for testimony. However, the judge ruled that the challenged evidence would not be admissible until its relevance was established at the time of its proffer.
Later in the trial, after the Government had rested its case, the defense moved in limine to determine admissibility of the test if the accused were to take the stand and deny any illegal drug use or if the defense were to call character witnesses. The military judge ruled that, in either instance, the results could be used by the prosecution in cross-examination but that extrinsic evidence of the results could not be used. Then, the defense announced that, because of this ruling, Sutton would not testify and the defense would call no character witnesses.
On appeal to the Court of Military Review, appellant challenged the military judge’s ruling. However, citing Luce v. United States, 469 U.S. 38, 105 S.Ct. 460, 83 L.Ed.2d 443 (1984), the majority of the court below concluded that it should not review this issue, because Sutton’s failure to take the stand or offer character witnesses prevented it from determining whether appellant had been prejudiced even if the judge had erred. In so reasoning, the majority acknowledged:
Luce does appear to represent a departure from prior military practice as discussed by the Court of Military Appeals in describing how issues such as those raised by the appellant’s motion in limine are preserved. See United States v. Cofield, 11 M.J. 422, 431 (C.M.A.1981). We note, however, that Luce has subsequently been cited favorably by that Court. United States v. Welch, 25 M.J. 23, 26 (C.M.A.1987). This suggests to us that the Court of Military Appeals accepts, at least in principle, the validity of the rule announced therein.
27 MJ at 581.
In dissenting from this position, Judge Blommers did not agree either that Luce applied or that, under these circumstances, an accused should be placed on the horns of a dilemma whether to testify and present character evidence in the face of the judge’s ruling or to risk losing appellate review of that ruling. As to this dilemma, he observed:
I believe there is no doubt but that an accused person’s right to testify and present other evidence on his or her behalf are fundamental and substantial trial rights. If there is a valid basis for use of impeachment or rebuttal evidence as a trial unfolds, and for tactical reasons the accused decides to waive those rights in order to keep such evidence from coming before the court, then so be it. Id. [Luce v. United States, supra ]. If there is no valid basis for the introduction of such evidence, then a rule that forces an accused to consider giving up those rights in light thereof offends my sense of justice.
27 MJ at 581. It is in relation to this latter point that Judge Blommers distinguished Luce:
There is nothing in that [Luce ] opinion that would lead me to conclude other than that the evidence the Government possessed was a properly documented prior conviction which fully met the procedural requirements for admissibility under Federal Rule of Evidence 609, assuming the defendant took the stand and “opened the door.” Thus the [valid basis for use of the challenged evidence under some circumstances] had already been satisfied.
27 MJ at 582.
II
A
In Luce, the defendant had contended on appeal that the trial court had erred “in [14]*14denying the motion in limine [to prohibit use of a prior conviction to impeach his credibility] without making an explicit finding [as required by Fed.R.Evid. 609(a)] that the probative value of the prior conviction outweighed its prejudicial effect.” 469 U.S. at 40, 105 S.Ct. at 462 (footnote omitted). However, six of the eight Justices who participated in the decision held “that to raise and preserve for review the claim of improper impeachment with a prior conviction, a defendant must testify.” 469 U.S. at 43, 105 S.Ct. at 464. Since Luce had not testified, the Court affirmed the refusal below to review the trial court’s in limine ruling.
In arriving at its conclusion, the Court noted that “[a] reviewing court is handicapped in any effort to rule on subtle evidentiary questions outside a factual context.” For instance, the Court observed that any number of factors might well subject the ruling to change during the course of the trial, depending on other factors, and that this fluidity of the admissibility ruling made harm to the accused from the ruling “wholly speculative.” Id. at 41, 105 S.Ct. at 463.
Other factors, as well, led the Court to its conclusion that, unless and until the defendant testifies, any ruling that his testimony may be impeached by evidence of a prior conviction should not be reviewed on appeal. Among these factors were the possibility that, even if the defendant does testify, the Government will forego impeaching him with the prior conviction because its case is strong or other means of impeachment are available; the uncertainty whether a defendant, despite his representation to the contrary, would in fact take the stand if the ruling were otherwise; and the recognition that, in testing for harmless error, “the appellate court could not logically term ‘harmless’ an error that presumptively kept the defendant from testifying,” which, in most cases, “would result in the windfall of automatic reversal.” Id. at 42, 105 S.Ct. at 463.
Justice Brennan, joined by Justice Marshall, concurred in the majority opinion because they understood its holding to be limited to reviewability of in limine rulings as to admissibility of prior convictions under Fed.R.Evid. 609(a), rather than extending to “deciding broader questions of appealability vel non of in limine rulings that do not involve Rule 609(a).” They expressly indicated that in cases
in which the determinative question turns on legal and not factual considerations, a requirement that the defendant actually testify at trial to preserve the admissibility issue for appeal might not necessarily be appropriate. The appellate court’s need to frame the question in a concrete factual context would be less acute, and the calculus of interests correspondingly different, than in the Rule 609(a) case the Court decides today.
Id. at 44, 105 S.Ct. at 464.
B
Use of motions in limine2 often avoids delays and interruptions in a trial, permits better scheduling of the presentation of evidence, and reduces the likelihood of mistrials. For reasons like those which gave rise to such motions in civilian courts, see Luce v. United States, supra at 41 n.4, 105 S.Ct. at 463 n. 4, motions in limine have been increasingly employed in trials by courts-martial. Authority for such motions is not expressly granted by the Rules of Evidence. However, it may be found in Article 39(a)(1) of the Uniform Code and has been inferred from the military judge’s inherent power to manage the trial, see Mil.R.Evid. 103(c) and 104(c), Manual for Courts-Martial, United States, 1984; and [15]*15these motions have been viewed by many judges as an effective tool for trial management.
As the Supreme Court pointed out in Luce, there are practical limitations to the use of motions in limine. Indeed, more than 3 years before Luce was decided, we recognized these limitations in United States v. Cofield, 11 MJ 422 (CMA 1981) —a case that, in many ways, was similar to Luce.
There, the defense had moved in limine for the military judge to rule that a summary court-martial conviction “could not be used ‘for purposes of impeachment of the defendant should he take the stand as well as for aggravation.’ ” Id. at 423. The basis for the motion was an offer of proof that Cofield had not received the advice of counsel at that court-martial and that, indeed, he had been advised that he had no right to such advice. From this offer of proof, defense counsel had argued that our earlier decision in United States v. Booker, 5 MJ 238 (CMA 1977), prohibited use of the record of the earlier court-martial either for impeachment or in aggravation. Ultimately, however, the military judge had ruled that the evidence could be used for both purposes.
Later in the trial, defense counsel had asked the judge to reconsider this ruling. He had explained that it was “critical” that Cofield testify. On the other hand, the defense realized that
“it is extremely damaging ah — for the defendant to be impeached with a prior conviction where he wasn’t represented by counsel and we would simply ask the Court to reconsider that ruling because it truly places the defense on the horns of a dilemma and it might — you know — clearly made [sic] a difference as to whether or not we can or cannot call the defendant to the stand.”
11 MJ at 425. Notwithstanding an expression of some sympathy for this “dilemma,” the military judge had adhered to his earlier ruling; and Cofield did not testify on the merits of his case.
In the Court of Military Eeview, Cofield had cited the military judge’s ruling as prejudicial error. Nonetheless, that court had “held in an unpublished opinion that, since the appellant had not testified on the merits, it would not speculate as to whether he had been prejudiced in some way.” Id. at 426.3
At the outset of our consideration of Cofield’s appeal from that decision, we observed:
These three separate questions can be identified as implicit in the present appeal: (a) Since he did not testify, is appellant entitled to review of the military judge’s ruling as to the admissibility of the summary court-martial conviction for impeachment purposes? (b) If he is entitled to review, was the judge’s ruling in error? and, (c) If error occurred, was appellant prejudiced thereby? A close relationship exists between these three issues. One reason for judicial reluctance to allow review under circumstances like those of the case at bar is the difficulty of determining somewhat in the abstract whether an erroneous ruling has been made. For example, even if we rule that usually a conviction by summary court-martial is inadmissible to impeach an accused after he has testified on direct examination, there may be extraordinary circumstances when such a conviction would be admissible — for example, if an accused uses inadmissibility as a sword, rather than a shield, by testifying on direct examination that he has never been convicted of an offense by any kind of court-martial. Similarly, the difficulty of establishing whether an erroneous ruling prejudiced an accused [16]*16who has not taken the stand undoubtedly has helped induce some courts to forestall the inquiry as to prejudice by simply refusing to review the ruling of the trial court unless the appellant has testified.
11 MJ at 426-27 (citations and footnote omitted).
Despite these concerns, our subsequent extensive analysis of the developing case-law in Federal courts revealed an “increasing willingness to undertake review under such circumstances.” Id. at 427 (footnote omitted). We observed, however, that
[djespite th[is] developing case law ..., appellate government counsel have strenuously urged that such a result would be inappropriate in cases like that at hand. They argue that, even if the appellant had testified, “it is quite possible that the summary court-martial conviction would have been admissible as rebuttal, rather than only as impeachment.” The Government also notes the procedural problems that derive from allowing review of such rulings on motions in limine — the most significant being uncertainty of whether the defendant is only seeking to establish a point for appeal and would not have testified even if the judge had ruled favorably on his motion.
Id. at 430 (footnote omitted).
Acknowledging the merit in this argument, we concluded that the military judge ought to have considerable discretion in determining whether to defer ruling on motions in limine, despite the benefits of such motions by way of advance planning. We suggested that the military judge, in exercising this discretion whether to defer, should consider such factors as the possibility of a mistrial if the challenged evidence is held to be inadmissible after the prosecution uses it in front of the members; waste of court members’ time while a motion is litigated in the midst of trial if a ruling is deferred until such time as the evidence is offered; the existence of some reasonable assurance that the ruling actually will be determinative of whether the accused will testify; and the fact that “many defendants are deterred from” exercising their substantial right to testify in their own behalf by the “fear that the prior convictions used to impeach their credibility will have a spillover effect in persuading the trier of fact of their guilt.” Id. at 431.
In the end, we noted that Cofield’s trial motion in limine, based on Booker, “presented a pure issue of law which could be readily ruled on.” Moreover, we found “nothing in the record which refutes or calls into question the defense counsel’s representation that Cofield would take the stand if the prior conviction were ruled inadmissible.” Accordingly, we concluded that this Court could “properly review the issue raised by [the military judge’s] ruling.” Id. at 431.
C
Since our decision in Cofield, the Rules for Courts-Martial have been promulgated in the Manual for Courts-Martial, United States, 1984. Those rules reflect in a number of places the continuing sensitivity to the need for the military judge to have utmost flexibility in managing the course of a trial and reflect, as well, our considerations in Cofield.
For example, RCM 102(b) makes clear that “[t]hese rules shall be construed to secure simplicity in procedure, fairness in administration, and the elimination of unjustifiable expense and delay.” RCM 801(a) charges that “[t]he military judge shall ... (3)[s]ubject to the code and this Manual, exercise reasonable control over the proceedings to promote the purposes of these rules and this Manual.” The Discussion of this provision notes that this rule, among other things, permits the military judge to determine “when, and in what order, motions will be litigated (see R.C.M. 905).”
This flexibility as to pretrial motions also is apparent in RCM 905(b), entitled “Pretrial motions.” That rule provides: “Any defense, objection, or request which is capable of determination without the trial of the general issue of guilt may be raised [17]*17before trial.” According to the Discussion of subsection 3 of this rule: “Questions concerning the admissibility of evidence on ... grounds [other than involuntary confessions and admissions, unlawful searches and seizures, and eyewitness identification] may be raised by objection at trial or by motions in limine.”
Perhaps most specific of all is RCM 906(b)(13), which lists a “[preliminary ruling on admissibility of evidence” as one of the matters that “may be requested by motion for appropriate relief.” As with the previously mentioned rules, the Discussion accompanying this provision leaves no doubt as to the intended flexibility of the rule:
A request for a preliminary ruling on admissibility is a request that certain matters which are ordinarily decided during trial of the general issue be resolved before they arise, outside the presence of members. The purpose of such a motion is to avoid the prejudice which may result from bringing inadmissible matters to the attention of court members.
Whether to rule on an evidentiary question before it arises during trial on the general issue is a matter within the discretion of the military judge. But see R.C.M. 905(b)(3) and (d); and Mil.R.Evid. 304(e)(2); 311(e)(2); 321(d)(2).
RCM 905(d), referenced in this Discussion, expressly provides for the flexibility in the discretion of the military judge that we found appropriate in Cofield:
A motion made before pleas are entered shall be determined before pleas are entered unless, if otherwise not prohibited by this Manual, the military judge for good cause orders that determination be deferred until trial of the general issue or after findings, but no such determination shall be deferred if a party’s right to review or appeal is adversely affected.
III
The Rules for Courts-Martial— which are promulgated by the President in the exercise of his authority under Article 36 of the Uniform Code of Military Justice, 10 USC § 836 — appear to have accepted the philosophy of Cofield. See, e.g., Drafters’ Analysis to RCM 906(b)(13), Manual, supra at A21-49. However, Luce, decided by the Supreme Court under the Federal Rules of Evidence, seems to reach a different result.4
Our decisions have sent mixed signals as to how this seeming conflict should be resolved. See United States v. Chambers, 29 MJ 76, 77 (CMA 1989); United States v. Gamble, 27 MJ 298, 307 (CMA 1988); United States v. Welch, 25 MJ 23, 26 (CMA 1987). Because appellant and his counsel reasonably relied on Cofield and the Rules for Courts-Martial, it would be unfair for us to refuse to review the military judge’s in limine ruling in this case. Therefore, we shall decide whether the military judge properly concluded that a “negative” urinalysis report could be used by the prosecution in cross-examining Sutton and his potential character witnesses.
However, we also need to provide guidance as to reviewability of in limine rulings in future cases. Under Cofield and the present Rules for Courts-Martial, a military judge has considerable discretion to defer rulings on in limine motions and thereby to avoid some of the problems which apparently concerned the Supreme Court in Luce. So far as I can discern, military judges have exercised their discretion properly, so there is no need to impose a condition that an accused testify as a prerequisite for appellate review of an in limine ruling on admissibility of a prior conviction or of other impeaching evidence. Thus, absent a decision by the President to [18]*18change the Rules for Courts-Martial, I am of the view that Luce does not affect courts-martial practice — other than to reemphasize factors that a military judge should consider in exercising his discretion.
My Brothers, however, are of a different view and would apply Luce in the same manner as it is to be applied in Federal criminal trials. The Court holds, therefore, that reviewability of in limine rulings in courts-martial tried after the date of this opinion will be controlled by Luce and by future decisions of the Supreme Court.5
IV
We now return to review this case under the law existing prior to the date of this opinion.
When, at the outset of the trial, defense counsel made a motion to suppress any evidence of the results of the “negative” RIA test, trial counsel stated that he could “foresee a number of uses that could be made of this evidence” — such as for showing Sutton’s knowledge or lack of mistake or for impeachment purposes. However, in light of the judge’s ruling that this evidence was not admissible until its relevance was established, the Government did not try to introduce it during the ease-in-chief.
When the Government rested and the defense made its motion in limine, the issue was whether the “negative” urinalysis could be used to cross-examine Sutton, if he testified, or any defense character witnesses. Cross-examination of appellant would have been based on Mil.R.Evid. 608(b) and 611(b). In cross-examining character witnesses, the Government would have relied on Mil.R.Evid. 405(a) and 611(b).
The use of the “negative” results in cross-examining defense character witnesses would have presented special problems. Presumably, trial counsel would have asked a character witness “do you know” or “have you heard” about the result of the urinalysis. If the prosecutor were only allowed to inquire whether the witness had learned that the accused had a “negative” test, the answer could not possibly help the Government’s cause. On the other hand, if the question were more specific and were phrased in terms of nanograms per milliliter, the answer might be misleading, unless extrinsic evidence were received to explain the test results; and reception of this extrinsic evidence would seem to violate Mil. R.Evid. 608(b). We conclude, therefore, that, even if the Government could have used the RIA test results to impeach Sutton if he had testified, it could not have used this evidence in the cross-examination of the defense character witnesses. Cf. United States v. Kitching, 23 MJ 601 (AFCMR 1986), pet. denied, 24 MJ 441 (1987).
A far more basic problem with the Government’s intended use of the test results was that such use would contravene Department of Defense Directives and Air Force regulations. United States v. Arguello, 29 MJ 198 (CMA 1989). As we made clear in Arguello:
The Supreme Court and this Court have also recognized that the Government is bound by its own regulations, especially when the regulation is one which confers a right or which benefits an individual. Cf. United States v. Caceres, 440 U.S. 741, 753, 99 S.Ct. 1465, 1472, 59 L.Ed.2d 733 (1979), and United States v. McGraner, supra [13 MJ 408 (CMA 1982)]. Here, trial counsel’s presentation of scientific evidence and his additional rebuttal argument that this negative test result was really positive violated such regulations. Cf. United States v. Pollard, 27 MJ 376 (CMA 1989).
29 MJ at 203.
We recognized in Arguello that the Department of Defense has established higher standards for administration of the [19]*19drug-testing program in order to avoid reporting of “false positive[s].” Id. at 204. Certainly such a policy choice was not irrational; and, this choice having been made, the regulations cannot be undercut by prosecutors seeking to prove that what the Government requires to be reported as “negative” is really “positive.” Moreover, to take a different position would raise constitutional issues because of the destruction of the urine specimen after the RIA screening test has yielded a “negative” result. See id. at 206; California v. Trombetta, 467 U.S. 479, 104 S.Ct. 2528, 81 L.Ed.2d 413 (1984).
Of course, our premise has been that the RIA screening test was done in accord with the procedures prescribed by the Department of Defense. The Government is not precluded from offering evidence that a urine specimen was diluted or improperly tested or that the chain of custody of the specimen was inadequate. See 29 MJ at 206. Indeed, if the test results were improperly analyzed and reported so that the results were reported as “negative” when under the applicable regulations they should have been positive, the Government may offer expert testimony to reinterpret the results. See United States v. Joyner, 29 MJ 209 (CMA 1989). However, the Government may not disregard its own regulations by seeking to prove in a trial by court-martial that results of an RIA screening test properly reported as “negative” should really be treated as “positive.”
In determining prejudice, we obviously labor under the disadvantage that we cannot be sure whether appellant would have testified in his own behalf or would have offered character witnesses during the trial on the merits or during the sentencing proceedings if the military judge’s ruling had been different.6 However, from our consideration of the record, it does appear very likely that Sutton could have called some character witnesses in his defense.7 In a case like this — where the Government relied solely on a positive urinalysis — such character evidence alone might have sufficed to “generate reasonable doubt” on the part of the court members. See United States v. Vandelinder, 20 MJ 41, 47 (CMA 1985).
As we read the record of trial, the military judge’s ruling as to admissibility of the test results had the practical effect of preventing the defense from offering any witnesses. That ruling contravened Department of Defense and Air Force regulations intended to safeguard the rights of servicemembers and to avoid conviction on unreliable evidence. Under these circumstances, we conclude that the requisite prejudice exists.
V
The decision of the United States Air Force Court of Military Review is reversed. The findings and sentence are set aside. The record of trial is returned to the Judge Advocate General of the Air Force. A rehearing may be ordered.