Peer-relationships are critically important for adolescent behavior, but how peer-friendship network composition and structure influence adolescent self-harm is less clear. The aim of this cross-sectional study was to explore the association between in-school peer-friendship networks, gender, and self-harm among inner-city adolescents. Participants were 2203 adolescents (mean age = 12.5 years, SD = 1.0; 53% girls) attending inner-city south London schools. Each adolescent nominated friends within their school year to construct sociometric peer-friendship networks and reported on lifetime self-harm. Mixed-effects logistic regression was used to estimate the effects of a comprehensive array of peer-network metrics on self-harm in the sample overall and by gender. Having friends who report self-harm, network over-integration (bridging, popularity), and social isolation (network under-integration) increased odds of self-harm, while sociality and high friendship group density reduced odds. Odds ratios did not vary by gender. The findings indicate that peer-network composition, particularly if friends self-harm, and over- and under-integration in wider peer-networks, may influence early adolescent self-harm, among both boys and girls.