Understanding the homogeneity of peer groups requires identification of peer groups and consideration of influence and selection processes. Few studies have identified adolescent peer groups, however, or examined how they become homogeneous. This study used social network analysis to identify peer groups (cliques), clique liaisons, and isolates among adolescents in 5 schools at 2 data collection rounds (N = 926). Cigarette smoking was the behavior of interest. Influence and selection contributed about equally to peer group smoking homogeneity. Most smokers were not peer group members, however, and selection provided more of an explanation than influence for why isolates smoke. The results suggest the importance of using social network analysis in studies of peer group influence and selection.