This proposal continues a 15-year research program that has rapidly accumulated evidence on how movies, other media and marketing influence youth smoking. The research program developed a novel assessment of exposure to behaviors in movies that involves content analysis of movies to ascertain the behavior of interest and special survey techniques to estimate population-based exposures to those behaviors in contemporary movies. This mature research program has collected 6 waves of data on a U.S. adolescent telephone cohort recruited in 2003 (original cohort, N = 6522), and two waves on a national telephone cohort of 15-23 year olds recruited in 2010 (ARRA cohort, N = 3341), with an additional wave planned. We aim to engage 8 PIs responsible for 11 cohort studies to create an aggregated dataset that includes all two-wave studies of movies and smoking (with a total of 53,284 baseline participants), in order to test the specificity of movie smoking effects. This collaborative team of investigators will meet monthly by Webex and annually in person to conduct a mega- meta-analysis that involve head-to-head comparisons of movie influence, exploring for example, how adding exposure to movie violence as a covariate affects the longitudinal association between movie smoking and smoking onset. Using a subset of the pooled dataset, we will test a novel theory of media influence that draws on our understanding of media appeals to swing voters-the idea that adolescents at intermediate risk for smoking (based on a traditional set of risk factors) are more susceptible to media influence than adolescents at very low or very high risk. With the ARRA cohort, we will compare the influence of entertainment media (movie and TV smoking) with that of tobacco marketing (direct mail receipt and tobacco website use) on smoking outcomes. We hypothesize that entertainment media primarily effects smoking onset, and marketing primarily effects progression of smoking among experimental smokers. This team is uniquely placed to conduct the proposed observational research and many of the co-investigators and collaborators are deeply involved in better understanding the science of media and marketing effects on behavior. Therefore, we expect the team to not only address the four specific aims delineated in the proposal but also advance additional aims and test them as well. This study is poised to greatly enhance our understanding of how entertainment media and marketing affect substance use behaviors.