Recent estimates indicate that a substantial percentage of individuals with autism spectrum disorder (ASD) are not cognitively impaired. Yet, outcomes for adults with ASD are poor. This suggests that interventions targeting the core symptoms of ASD need to be lifelong. This project will compare the effectiveness of two variants of one of the only empirically supported social skills interventions for adolescents with ASD, the PEERS program. In a well-characterized sample of 48 cognitively-able adolescents with ASD, change in selfand parent-reported social functioning and mental health will be compared among participants who complete either (1) the traditional PEERS program, (2) a peer-mediated variant of the PEERS program (i.e., PEERS with typical peers), or (3) a delayed treatment control (DTC) group. The short-term and long-term generalizability of skills learned during the PEERS programs will also be examined by quantifying adolescents’ social networks within their classrooms. It is predicted that adolescents who complete PEERS with peers will demonstrate (1) more improvement in social skills and mental health and (2) more reciprocated friendships and social network centrality than adolescents in the other two groups. Application of these findings may provide new guidelines for implementation of PEERS and other interventions that will improve social integration of adolescents with ASD in real-world settings.