Early adolescence may be an opportune time for violence prevention, particularly if shifts in risk patterns and the importance of peer influence are considered. An important question is whether to target high‐risk students or the entire population. Thirty‐seven schools were randomized to four conditions: universal classroom intervention; selective family intervention; combined interventions; or control to test effects on a high‐risk sample (N = 1,805) of sixth graders targeted due to elevated aggression and social influence. Intent‐to‐treat and dosage‐weighted growth comparisons through two years of postintervention revealed selective intervention benefits for physical aggression, aggressive strategies, and targeted family characteristics. Universal and combined interventions affected valuing academic achievement. Implications for middle school prevention, particularly selective targeting of socially influential high‐risk youth, are discussed.