Sexual health problems cause negative externalities from contagious diseases and public expenditure burdens from teenage pregnancies. In a randomized evaluation, we find that an online sexual health education course in Colombia leads to significant impacts on knowledge and attitudes but no impact on selfreported behavior, on average; although fewer STIs are reported for baseline sexually active females. To go beyond self-reported measures, we provide condom vouchers six months after the course to both treatment and control groups and estimate a 9 percentage point treatment effect (52% increase) on the likelihood of redemption. Using knowledge of friendship networks, we document a strong social reinforcement effect: the impacts of the course intensify when a larger fraction of a student’s friends is also treated. In particular, when full sets of friends are treated we find significant reductions in sexually active, frequency of sex, and number of partners. Throughout the analysis we fail to find evidence of cross-classroom spillovers.