We examined the efficacy and mediation of Being Responsible for Ourselves (BRO), an HIV/STI risk-reduction intervention for African American men who have sex with men (MSM), the population with the highest HIV-diagnosis rate in the US. We randomized African American MSM to one of two interventions: BRO HIV/STI risk-reduction, targeting condom use; or attention-matched control, targeting physical activity and healthy diet. The interventions were based on social cognitive theory, the reasoned-action approach, and qualitative research. Men reporting anal intercourse with other men in the past 90 days were eligible and completed pre-intervention, immediately post-intervention, and 6 and 12 months post-intervention surveys. Of 595 participants, 503 (85 %) completed the 12-month follow-up. Generalized-estimating-equations analysis indicated that, compared with the attention-matched control intervention, the BRO intervention did not increase consistent condom use averaged over the 6- and 12-month follow-ups, which was the primary outcome. Although BRO did not affect the proportion of condom-protected intercourse acts, unprotected sexual intercourse, multiple partners, or insertive anal intercourse, it did reduce receptive anal intercourse compared with the control, a behavior linked to incident HIV infection. Mediation analysis using the product-of-coefficients approach revealed that although BRO increased seven of nine theoretical constructs it was designed to affect, it increased only one of three theoretical constructs that predicted consistent condom use: condom-use impulse-control self-efficacy. Thus, BRO indirectly increased consistent condom use through condom-use impulse-control self-efficacy. In conclusion, although BRO increased several theoretical constructs, most of those constructs did not predict consistent condom use; hence, the intervention did not increase it. Theoretical constructs that interventions should target to increase African American MSM’s condom use are discussed.