Appellant was tried by a general court-martial in September and October 1991 at Twentynine Palms, California. Contrary to his pleas, he was found guilty of having carnal knowledge with a 14-year-old girl, in violation of Article 120, Uniform Code of Military Justice, 10 USC § 920. He was sentenced by officer members to a bad-conduct discharge, confinement for 9 months, total forfeitures, and reduction to the lowest enlisted pay grade. The convening authority approved the adjudged sentence on April 15, 1992. The Court of Criminal Appeals affirmed on June 19, 1995, in an unpublished opinion.
On October 31, 1995, this Court granted review on the following issue:
I
WHETHER THE NAVY-MARINE CORPS COURT OF CRIMINAL APPEALS ABUSED ITS DISCRETION BECAUSE IT HELD THAT THE MILITARY JUDGE’S ERRONEOUS ADMISSION OF DNA EVIDENCE WAS HARMLESS ERROR.
We hold that the Court of Criminal Appeals did not err in holding that admission of DNA (deoxyribonucleic acid) evidence was harmless error, if error at all. See United States v. Weeks, 20 MJ 22 (CMA 1985); see generally United States v. Youngberg, 43 MJ 379 (1995).
*495Appellant was found guilty of having carnal knowledge with HP, a 14-year-old girl with subnormal intelligence. At the time, he was a 31-year-old gunnery sergeant in the Marine Corps. Several witnesses testified that they observed appellant and HP behaving in a manner inconsistent with this disparity in their ages. HP told a number of persons that appellant was her boyfriend and had given her a number of gifts, including a telephone credit card so she could call him. At trial, HP testified that she and appellant began a sexual relationship in the spring of 1990, which continued until October of that year. She also stated that, during this time, they engaged in sexual intercourse over 50 times.
In November of 1990, HP was found to be pregnant. The fetus was aborted. The doctor who performed the procedure testified that HP was about 8 weeks pregnant. At the time relevant to conception, HP had stayed with appellant for about 2 weeks because of the death of her grandmother. Moreover, HP testified that appellant was the only person with whom she had engaged in sexual intercourse. Tissue samples were taken from the fetus, and blood was taken from HP. Based on the allegations of sexual contact between appellant and HP, a warrant was obtained for a sample of appellant’s blood.
The court below described the challenged DNA evidence in this case as follows:
In summary, that evidence was developed from testing and comparing blood samples from HP and the appellant and a tissue sample from the aborted fetus. Four separate tests were conducted using different “probes” to test different segments of DNA strands from each of the three samples. Based on a comparison of the results of the four tests (referred to as “genotyping”), the appellant could not be excluded as the possible father. Any one of the four tests could have conclusively determined that it was impossible for the appellant to be the father, but none of the tests could conclusively determine that the appellant was the father. In other words, each test could only reveal whether it was possible or impossible for the appellant to have been the father. Consequently, the test results, alone, were of limited probative value.
The Government’s expert was then permitted to make statistical comparisons with DNA data bases expressed as what the chances would be that any random, unrelated male in the data base could be excluded as the possible father. The data bases used included one the expert stated was derived from testing 150 to 175 “Utah Caucasians;” one of “Texas Hispanics” developed by the Federal Bureau of Investigation [FBI], which the expert admitted in cross-examination was less than 100 and may have been as low as 43 individuals; a larger FBI data base of “Texas Hispanics” that was constructed from about 180 to 200 individuals; and a third FBI data base for “black men.” From these comparisons, the Government expert testified that: (1) of the Utah group, from 1 in 1,000 to k of 100,000 could not be excluded as a possible father, (2) of the first Texas group 1 or 2 out of 100 could not be excluded; (3) of the second, larger Texas group 7 in 1,000 could not be excludedand, (tí of the black males, 5 or 6 in 1,000 could not be excluded.
Unpub. op. at 6-7 (emphasis added).
The Court of Criminal Appeals held that any error in admitting the DNA statistical evidence in this case was harmless. See Youngberg, 43 MJ at 387 n. 10; United States v. Davis, 40 F.3d 1069, 1075 (10th Cir.1994) (statistical portion of DNA evidence a matter of weight). In particular, it said:
The fact that HP was pregnant was uncontroverted, and so was her testimony that the appellant was the only male with whom she had engaged in sexual intercourse during the months preceding her pregnancy. Not only was the appellant seen engaged in inappropriate and suggestive behavior with HP during those months, but no evidence of similar contact — or even of any contact — with another sexually mature male as a possible cause of HP’s pregnancy was offered.
*496As a starting point, we note that the granted issue implies that this Court reviews no-prejudice determinations by the Courts of Criminal Appeals using an abuse-of-discretion standard. No authority is cited for this proposition, and no argument is presented to justify use of this standard with respect to evidentiary rulings found to be erroneous by the Courts of Criminal Appeals. See S. Childress and M. Davis, Federal Standards of Review § 4.03 at 4-34 — 35 (prejudice analysis requires de novo review by appellate court) and § 11.16 at 11-9 (Discretion and Confusion with Harmlessness) (2d ed.1992). We hold that a de-novo-review standard to assess prejudice from erroneous evidentiary rulings is not only more appropriate but the one required by Article 59(a), UCMJ, 10 USC § 859(a), and Mil.R.Evid. 103(a), Manual for Courts-Martial, United States, 1984. See United States v. Grooters, 39 MJ 269, 273 (CMA 1994); United States v. Brabant, 29 MJ 259, 263 (1989); see generally Brecht v. Abrahamson, 507 U.S. 619, 636, 638, 113 S.Ct. 1710, 1722, 123 L.Ed.2d 353 (1993); Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946).
Turning now to this task, we note that the precise question before the members in this case was whether appellant engaged in sexual intercourse with HP. Art. 120(b). HP testified that they did engage in this act over 50 times. Medical testimony was also introduced which established that she had been pregnant, a physical condition usually caused only by sexual intercourse. Finally, the DNA evidence offered by the prosecution showed only that appellant could not be ruled out as the father of the aborted fetus. The statistical portion of this testimony was offered to show that such a conclusion (possible paternity) would be rare in certain test groups.
In our view, the above-noted DNA evidence clearly had no substantial influence on the members’ findings in this ease. See United States v. Tome, 61 F.3d 1446, 1455 (10th Cir.1995). Scientific evidence that appellant could not be ruled out as the father hardly established an ironclad case of his guilt. See United States v. Walker, 42 MJ 67, 74 (1995). Logically, the probative value of statistical evidence establishing the rarity of such an inconclusive determination would also not be great, particularly in this case where the prosecution failed to particularly place appellant in one of the test groups. Id. Finally, the import of this somewhat extraneous scientific evidence pales in comparison to other circumstantial evidence of appellant’s guilt in this case, which was noted in great detail by the court below. See United States v. Ray, 26 MJ 468, 472 (CMA 1988); United States v. Duncan, 42 F.3d 97, 103 (2d Cir. 1994). Admission of the above scientific evidence was harmless error, if error at all. See United States v. Youngberg and United States v. Davis, both supra.
The decision of the United States Navy-Marine Corps Court of Criminal Appeals is affirmed.