Background Few, if any, studies have differentiated patterns of alcohol and marijuana use as predictors of sexual risk among adolescents. The objective was to compare recent use of alcohol and marijuana (A+M), alcohol only (AO) and marijuana only (MO) as predictors of condom use, pregnancy and sexually transmitted infections (STIs) during 18 months of follow-up among African-American female adolescents.

Methods The analytic sample included 213 African-American females (14–20 years) assigned to the control arm of an HIV prevention trial. Participants completed audio computer-assisted self-interviews at baseline and every 6 months for 18 months. At each assessment, a pregnancy test was conducted and specimens were collected and assayed for Trichomonas vaginalis, Chlamydia trachomatis and Neisseria gonorrhoeae using DNA amplification. Generalized estimating equations compared proportion condom-protected sex acts in the past 3 months, pregnancy and STI acquisition during follow-up among participants who in the 90 days prior to baseline assessment reported using A+M, AO and MO but no other illegal substances. Multivariable models controlled for group differences at baseline and corresponding baseline measure of the outcome.

Results Of 182 (85%) participants with follow-up data, 43% (n = 79) reported recent use of A+M, 39% (n = 71) AO and 17% (n = 32) MO. No group differences in retention or proportion condom-protected sex acts during follow-up were observed. There were no significant differences in STI acquisition or pregnancy among MO and AO users. Relative to AO users, A+M users were more likely to become pregnant (AOR: 2.5, 95% CI: 1.2, 5.3). Relative to MO users, A+M users were more likely to acquire an STI (AOR: 2.5, 95% CI: 1.0, 6.2) and become pregnant (AOR: 3.7, 95% CI: 1.1, 12.3).

Conclusion STI/HIV and pregnancy prevention programmes serving African-American female adolescents may benefit by intensifying services for recent A+M users and addressing the role of substance use on sexual risk.