OPINION OF THE COURT
BRESLIN, Judge:The appellant was convicted, in accordance with his pleas, of two specifications alleging wrongful use of cocaine, in violation of Article 112a, UCMJ, 10 U.S.C. § 912a. He was also charged with desertion, in violation of Article 85, UCMJ, 10 U.S.C. § 885, but was found guilty, in accordance with his pleas, of the lesser included offense of absence without leave, in violation of Article 86, UCMJ, 10 U.S.C. § 886. The sentence adjudged and approved included a bad-conduct discharge, confinement for 5 months, and forfeiture of $500.00 pay per month for 5 months. The appellant now avers that his due process rights were violated because the government did not disclose evidence before trial that might have impeached the reliability of the laboratory reports showing the appellant’s ingestion of cocaine. The appellant requests either that we set aside the findings of guilt to the wrongful use of cocaine and reassess the sentence, or that we order a hearing under United States v. DuBay, 37 C.M.R. 411, 1967 WL 4276 (1967), to determine what the government knew about the evidence in issue. Finding no error that materially prejudices the appellant’s substantial rights, we affirm.
Facts
On 20 July 1999, the appellant was selected at random to provide a urine sample as part of the Air Force Drug Testing Program at Vandenberg Air Force Base (AFB), California. The Air Force’s drug testing laboratory at Brooks AFB, Texas, analyzed the appellant’s specimen, and reported that it was positive for the metabolite of cocaine at 56,717 ng/ml, substantially above the Department of Defense’s cut-off of 100 ng/ml. The appellant’s commander preferred charges against the appellant on 7 September 1999, alleging wrongful use of cocaine.
Also on 7 September 1999, the appellant provided another mane sample for testing as part of an inspection of his entire unit. The second specimen was tested as before. The drug testing laboratory tested the second specimen as before, and reported it positive for the metabolite of cocaine at 951 ng/ml. Thereafter, the appellant’s commander preferred an additional charge alleging a second wrongful use of cocaine.
Trial was scheduled for 26 October 1999. The appellant did not appeal' at trial, however. He left Vandenberg AFB without au*527thority on 25 October 1999, and remained absent without leave (AWOL) until he reported to his commander on 2 November 1999. Thereafter, the appellant’s commander preferred a second additional charge alleging desertion. At trial, the appellant was found guilty in accordance with his pleas, as noted above.
While preparing post-trial clemency submissions, trial defense counsel obtained a copy of a report of investigation from the drug testing laboratory that cast doubt on the forensic integrity of urinalysis samples tested by Mr. Alexander Hatzis, a laboratory technician who had performed part of the testing of both of the appellant’s urine specimens. The report was dated 28 January 2000, and written by the drug testing laboratory’s Quality Assurance Oversight Office concerning a special audit of the work of Mr. Hatzis. The report concluded that Mr. Hatzis consistently and regularly violated standard operating procedures and accepted forensic practices regarding forensic documents, and that, while the studied test results were analytically sound, they had been “forensically compromised.” The report listed as attachments numerous separate documents, including a 5 November 1999 letter suspending/decertifying Mr. Hatzis from certain laboratory testing, a 19 November 1999 letter restricting him from access to the GC/MS laboratory, and a 29 November 1999 letter restricting him from access to the Investigations Room.
In his post-trial submissions, trial defense counsel maintained the government had failed to properly provide discovery by failing to disclose the adverse information regarding the drug testing laboratory’s procedures. Trial defense counsel noted the defense request for discovery submitted on 31 August 1999 (actually before the date of the first charge), which requested evidence of an exculpatory nature, evidence tending to negate the guilt of the appellant, or evidence of a derogatory nature concerning the drug testing laboratory. The appellant’s defense counsel asked the convening authority to substitute an administrative discharge for the court-martial sentence due to the alleged government violation of discovery rules.
Acting upon the recommendation of the acting staff judge advocate, the convening authority did not grant relief. He approved the findings and sentence as adjudged.
On appeal, the defense renews its argument that the government failed to disclose evidence material to the preparation of the defense. The appellant avers that, if the impeachment evidence of Mr. Hatzis had been disclosed, there would have been a high probability of a different result at trial. We ordered production of the relevant documentary evidence in order to review its probative value and consider its probable effect on the outcome of the case. See United States v. Dixon, 8 M.J. 149 (C.M.A.1979).
Law
The starting point for discovery matters is Article 46, UCMJ, 10 U.S.C. § 846, which provides: “The trial counsel, the defense counsel, and the court-martial shall have equal opportunity to obtain witnesses and other evidence in accordance with such regulations as the President may prescribe.” Acting pursuant to this delegation, the President promulgated Rules for Courts Martial (R.C.M.) 701 and 703. R.C.M. 701(a)(2)(A) requires the government to permit the defense, upon request, to inspect “[a]ny books, papers, documents ... or copies or portions thereof, which are within the possession, custody, or control of military authorities, and which are material to the preparation of the defense____” Subsection (B) of that rule provides a similar right to inspect:
Any results or reports of ... scientific tests or experiments, or copies thereof, which are within the possession, custody, or control of military authorities, the existence of which is known or by the exercise of due diligence may become known to the trial counsel, and which are material to the preparation of the defense____
R.C.M. 701(a)(2)(B) (emphasis added). R.C.M. 701(a)(6) sets out specific obligations with respect to evidence favorable to the defense:
The trial counsel shall, as soon as practicable, disclose to the defense the existence of evidence known to the trial counsel which reasonably tends to:
*528(A) Negate the guilt of the accused of an offense charged;
(B) Reduce the degree of guilt of the accused of an offense charged; or
(C) Reduce the punishment.
Finally, both sides have the continuing duty to provide additional evidence, and material requested or required to be produced, discovered before or during the court-martial. R.C.M. 701(d).
These Rules for Courts-Martial are consistent with the Supreme Court’s decision in Brady v. Maryland, 373 U.S. 83, 87, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), finding a violation of due process when the government withholds information requested by the defense that is material to the issue of guilt or punishment. It makes no difference whether the information is exculpatory evidence or impeachment evidence. United States v. Bagley, 473 U.S. 667, 676, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985). The key question is whether the evidence was “material to the preparation of the defense.” Id. “Impeachment evidence ... can obviously be material evidence at a criminal trial.” United States v. Watson, 31 M.J. 49, 54-55 (C.M.A.1990). See Giglio v. United States, 405 U.S. 150, 154, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972); United States v. Morris, 52 M.J. 193, 197 (1999).
R.C.M. 701(a)(2) places upon the trial counsel the duty of finding discoverable information. “[T]he individual prosecutor has a duty to learn of any favorable evidence known to the others acting on the government’s behalf in the case, including the police.” Kyles v. Whitley, 514 U.S. 419, 437, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995). The scope of the trial counsel’s inquiry, beyond his or her own files, depends upon the nature of the request and the trial counsel’s relationship to the holder of the information. United States v. Williams, 50 M.J. 436, 441 (1999).
If information is withheld improperly, the test for prejudicial error is whether there “is a ‘reasonable probability’ of a different result” had the suppressed evidence been disclosed to the defense. Kyles, 514 U.S. at 434, 115 S.Ct. 1555; Bagley, 473 U.S. at 682, 105 S.Ct. 3375; Williams, 50 M.J. at 440. “A ‘reasonable probability’ is a probability sufficient to undermine confidence in the outcome.” Bagley, 473 U.S. at 682, 105 S.Ct. 3375; Morris, 52 M.J. at 198. The defense need not demonstrate that “the suppressed evidence would have resulted” in acquittal. Kyles, 514 U.S. at 434, 115 S.Ct. 1555.
It is arguable that, where an accused makes a specific request for information and the government provides an incomplete response, the defense is doubly prejudiced, so that a higher standard of review is appropriate. United States v. Hart, 29 M.J. 407 (C.M.A.1990); United States v. Eshalomi, 23 M.J. 12 (C.M.A.1986). An incomplete response not only deprives the accused of certain information but may also have the effect of representing to the defense that the evidence does not exist, perhaps causing the defense to abandon lines of investigation or trial strategies it might otherwise have pursued. However, the Supreme Court specifically rejected a higher standard of review in such cases, requiring a reviewing court to. assess the possibility that such effect might have occurred in light of the “totality of the circumstances” and with an awareness of the difficulty of reconstructing the likely course of action had the defense not been misled. Bagley, 473 U.S. at 682-83, 105 S.Ct. 3375. See also United States v. Green, 37 M.J. 88 (C.M.A.1993).
Analysis
The appellant argues that the government erred in not providing the trial defense counsel with evidence of Mr. Hatzis’ suspension, decertification from testing and ultimate transfer. The government responds, in part, that the report was not written until after the appellant’s trial and that the government has no duty to create records to satisfy discovery demands. That is true. Kissinger v. Reporters Committee for Freedom of the Press, 445 U.S. 136, 100 S.Ct. 960, 63 L.Ed.2d 267 (1980); United States v. Birbeck, 35 M.J. 519, 522 (A.F.C.M.R.1992). However, from the record before us, it appears some responsive documents existed before the appellant’s trial, including letters ordering Mr. Hatzis’ suspension and decertification from testing.
*529The government argues that the failure to disclose was not in bad faith, since trial counsel was unaware of the additional information. The trial defense counsel requested discovery early in the case (indeed, before preferral of charges). The government responded to the discovery request on 20 September 1999, and provided some additional information thereafter. The appellant’s AWOL on the eve of trial delayed the case by nearly 2 months, during which time the information in question came to light. Most likely, after preferral of the third charge, all parties anticipated a guilty plea, so neither side made an effort to update the discovery information. We find no indication that the trial counsel acted in bad faith; rather it was an unexpected delay caused, in part, by the appellant’s AWOL which made the previously provided information outdated.
Nonetheless, R.C.M. 701(a)(2) places the burden on the government to discover and disclose to the defense information material to the preparation of the defense. With regard to books, papers, and documents, the trial counsel’s responsibility extends to those “which are within the possession and control of military authorities____” R.C.M. 701(a)(2)(A). For “results or reports of ... scientific tests or experiments,” the trial counsel must disclose those known to trial counsel, or those which, “by the exercise of due diligence may become known to the trial counsel....” R.C.M. 701(a)(2)(B).
Given that it is the prosecutor’s responsibility to provide discovery, it does not matter whether the failure to disclose is in bad faith or in good faith. Kyles, 514 U.S. at 437-38, 115 S.Ct. 1555; Giglio, 405 U.S. at 153, 92 S.Ct. 763. The prudent prosecutor should take steps to make sure that necessary information is collected properly, and that it is still current and complete after a delay.
We find that the evidence that Mr. Hatzis had been suspended and decertified from testing because of forensic discrepancies was discoverable under R.C.M. 701(a)(2). It is unnecessary to debate whether this was “exculpatory evidence,” “impeachment evidence” or simply more documents relating to the accuracy of the government’s scientific tests. Bagley, 473 U.S. at 676, 105 S.Ct. 3375. The information was certainly relevant to the reliability of the scientific tests that formed the basis of the government’s evidence regarding the drug offenses. Since the evidence was discoverable and it was the government’s obligation to obtain it, we find error in failing to disclose it to the trial defense counsel.
Having found error in failing to disclose this evidence, we must review the totality of the evidence to determine, “if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. A ‘reasonable probability’ is a probability sufficient to undermine confidence in the outcome.” Pennsylvania v. Ritchie, 480 U.S. 39, 57, 107 S.Ct. 989, 94 L.Ed.2d 40 (1987) (quoting Bagley, 473 U.S. at 682, 105 S.Ct. 3375).
Determining whether disclosure of suppressed evidence probably would have resulted in a different outcome depends upon the unique facts of each case. The failure to disclose impeachment evidence may be sufficient to require reversal, if it involves a key witness or issue. See Kyles, 514 U.S. at 441, 115 S.Ct. 1555 (in light of all the evidence disclosure of inconsistent witness statements and exculpatory police reports would have made a different result reasonably probable); Giglio, 405 U.S. at 154-55, 92 S.Ct. 763 (failure to disclose promise of leniency to prosecution’s primary witness required reversal); Eshalomi, 23 M.J. at 28 (outcome probably would have been different if prosecution had properly produced victim’s later, inconsistent statement).
At other times, failure to disclose impeachment evidence or exculpatory evidence has not required reversal. See Strickler v. Greene, 527 U.S. 263, 294, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999) (petitioner would have been convicted of capital murder and sentenced to death even if impeachment evidence concerning key witness had been disclosed); Wood v. Bartholomew, 516 U.S. 1, 116 S.Ct. 7, 133 L.Ed.2d 1 (1995) (disclosure of polygraph results would not create reasonable probability of a different result); Brady, 373 U.S. at 90, 83 S.Ct. 1194 (failure to disclose statement of co-conspirator confessing to actual killing did not require reversal *530of verdict); Morris, 52 M.J. at 198 (disclosure of victim’s medical and counseling records would not have produced a different outcome); United States v. Stone, 40 M.J. 420, 424 (C.M.A.1994) (failure to disclose impeachment evidence about government witness was harmless); Watson, 31 M.J. at 54-55 (failure to disclose victim’s claim against government for payment of $5,000 for injuries did not cast reasonable doubt on the proceedings); Green, 37 M.J. at 90 (no reasonable probability that the outcome of trial would be different if prosecutor had disclosed records of nonjudicial punishment usable as impeachment of prosecution witness); Hart, 29 M.J. 407 (C.M.A.1990) (no reasonable likelihood that failure to disclose results of scientific analysis of evidence would have affected the findings).
Turning to the facts in this case, we find that the appellant’s urine samples tested positive for cocaine on two separate occasions. The drug testing laboratory analyzed each sample twice using immunoassay procedures. The confirmation testing was done through Gas Chromatography/Mass Spectrometry (GC/MS) analysis. Mr. Hatzis performed the first GC/MS test on the appellant’s first sample. However, the results were not acceptable due to “column overload,” caused by the high concentration of drug in the appellant’s sample. The sample was then diluted and another technician completed the analysis, which was positive for the metabolite of cocaine at a concentration over 500 times the cut-off level. Mr. Hatzis was not involved in the GC/MS analysis that resulted in the positive confirmatory test of the appellant’s first sample.
The appellant’s second urine sample was also tested twice using immunoassay procedures, and confirmed positive for the metabolite of cocaine using the GC/MS procedures. Mr. Hatzis was involved in the GC/MS testing of this batch of samples.
The drug testing laboratory’s special audit report indicated Mr. Hatzis was suspended and decertified for failing to follow standard operating procedures at the drug testing laboratory. Specifically, he did not properly maintain documentation of the GC/MS testing performed, in violation of forensic standards. The GC/MS testing is computer-controlled, so that each step in the testing process generates documents identified by date and time, showing the process of the testing. It appears that portions of the test procedures performed by Mr. Hatzis, such as the injection of water blanks or standards, failed to achieve desired results. Rather than including these frustrated tests in the reports, he discarded the documents and entered inaccurate information in the logs to cover the resultant time gaps in the testing steps. The audit report speculates that Mr. Hatzis did this to make the test runs appear “cleaner,” which would, in turn, reflect more positively on his job performance.
The audit reviewed all the testing performed by Mr. Hatzis from July 1999 through 5 November 1999. This period includes the tests Mr. Hatzis performed on the appellant’s urine specimens. The audit reported several specimens where the computer records showed gaps in the testing process. The report concluded that the “various discrepancies did not impact the analytical validity of the test results. Valid standards, water blanks and low controls existed in each case.” In sum, the audit concluded that the test results were correct, but the paperwork was not maintained properly in some instances. Significantly, there were no discrepancies noted concerning Mr. Hatzis’ handling of the appellant’s samples.
Considering all the evidence in this case, and assuming that the evidence at issue had been properly disclosed to the trial defense counsel, we nonetheless conclude that it would not have resulted in a different outcome. We do not believe this evidence would have made the appellant change his mind about whether to enter guilty pleas — there already exists a substantial amount of information which a diligent defense counsel can use to attack the testing procedures at the drug testing laboratory. We note that Mr. Hatzis did not conduct the final, confirmatory testing for the appellant’s first sample, which resulted in a concentration over 500 times the DoD cut-off level. It is also very significant that surprise inspections of the appellant’s urine tested positive, not once, but *531twice, for the same illegal substance. The audit report could have been used at trial to attack the reliability of the laboratory testing procedures. However, the audit also indicated that experts looked again at the tests in this case and found no errors. If this evidence had been disclosed before trial, we are confident the appellant would have pled guilty nonetheless. In any event, we find that even if the appellant pled not guilty and used this evidence to good effect, it would not have resulted in a reasonable probability of a different outcome.
This is a close question, and the disposition of this case depends upon its specific facts. Under slightly different circumstances the failure to provide this evidence could result in reversal for a violation of the appellant’s due process rights. Government counsel — at all levels — must be vigilant to make sure such evidence is disclosed to the defense before trial.
The findings are correct in law and fact, and no error prejudicial to the substantial rights of the appellant occurred. Accordingly, the findings and sentence are
AFFIRMED.