DescriptionResearch has shown that school-based prevention programs can prevent substance misuse and promote graduation. Few studies have investigated high school-based programs, their long-term effects, and their intervening processes. Prosocial bonding to an entity outside of oneself has been theoretically (Finn, 1989; Hirschi, 1969) and empirically (McNeely & Falci, 2004) associated with positive behavioral outcomes for youth and rarely studied as a mediator of school-based prevention program effects. The current study examined the impact of Peer Group Connection (PGC) on students’ marijuana use and high school graduation. This study also examined school belonging as a mediator and acculturation as a moderator of potential program effects. In 2005, 268 students (92% Latino) were randomized to the control (n = 175) or program (n = 93) condition. Trained upperclassmen delivered weekly forty-minute, manualized sessions to ninth grade students. Latent growth curve analysis, using baseline, post-test, one-year follow-up, and two-year follow-up data showed an interaction. Specifically, the more acculturated PGC participants were more likely to graduate than similar controls (84.6% vs. 60.3%, respectively; Pseudo R-square = 9%). No evidence was found that school belonging mediated this effect. No intervention effect was observed for marijuana use. Thus, there is evidence that PGC benefited the Latino students who were most at risk for school dropout in this sample—those who were most acculturated.