Prevailing social network frameworks examine the association between peerties and behaviors like smoking, but the role of social isolates is poorlyunderstood. Some theories predict isolated adolescents are protected frompeer influence that increases smoking, while others suggest isolates aremore likely to initiate smoking because they lack social control providedby peer friendships. Building on a growing literature that seeks to explainthese contradictions by moving beyond a homogeneous understanding ofisolation, we identify the relationships between smoking and three distinctdimensions of isolation: avoided (adolescents who do not receive ties),withdrawn (adolescents who do not send ties), and externally oriented(adolescents who claim close out-of-grade friends). We examine thecoevolutionary effects of these dimensions and cigarette smoking using anautoregressive latent trajectory model (ALT) with PROSPER Peers, a unique,longitudinal networks dataset. These data include students (47% male and86% White) from rural Iowa and Pennsylvania, ranging successively fromgrades 6-12 in eight waves of data. As a robustness check, we use astochastic actor-oriented model (SAOM) to compare to results from the ALT.We find avoided isolation and external orientation are associated withdecreased successive smoking in high school, while smoking increasessubsequent isolation along all three dimensions, with particularly strongeffects on withdrawn isolation.