OBJECTIVE: To determine whether the effect of movie exposure to smoking on adolescent smoking onset is mediated through increased affiliation with peers who smoke. DESIGN: A longitudinal study was conducted with a sample of 5th- 8th graders; persons who were nonsmokers at the baseline assessment (N = 2,614) were followed up 18 months later. Movie exposure to smoking cues was assessed at baseline with a rigorous coding procedure. MAIN OUTCOME MEASURE: A school-based survey and follow-up telephone interview determined whether the participant smoked cigarettes. RESULTS: Longitudinal structural modeling analysis indicated movie-smoking exposure was related to smoking onset both through an indirect effect involving increased affiliation with peer smokers and through a direct effect. The analysis controlled for demographics, parenting style, rebelliousness and sensation seeking, school performance, parental smoking, and sibling smoking; several of these variables also had mediated or direct effects to smoking onset. CONCLUSION: The effect of movie exposure on adolescent smoking onset is attributable in part to a social mechanism. Implications of media effects for prevention are discussed. (PsycINFO Database Record (c) 2007 APA, all rights reserved).