Adolescent girls and young women (AGYW) in South Africa experience contextual barriers to HIV risk reduction including incomplete schooling, unintended pregnancy, substance use, and gender-based violence. A cluster randomised trial in Cape Town allocated 24 Black and Coloured communities to a gender-focused HIV risk-reduction education; heavy alcohol intervention or HIV testing, with 500 AGYW total enrolled. We evaluated use; condom use; intervention efficacy by comparing mean differences overall, by methamphetamine use community population group (Black and Coloured) and among those with structural barriers based on neighbourhood, education, and employment (n = 406). Both arms reported reductions in alcohol, cannabis, and condomless sex, with no intervention efficacy overall. Among AGYW with barriers, intervention participants reported fewer days of methamphetamine use at 6 months (t(210) = 2·08,
p
= ·04). In population group analysis, we found intervention effects on alcohol and sexual communication. Intervention participants in Black communities had fewer days of alcohol use at 12 months (t(136) = 2·10,
p
= ·04). Sexual discussion (t(147) = −2·47,
p
= ·02) and condom negotiation (t (146) = −2·51,
p
= ·01) increased for intervention participants at 12 months in Coloured communities. Gender-focused interventions must address population group differences and intersecting barriers to decrease substance use and increase education, skills, and sexual health protection.