A recent investigation by Rosenberg and Rosenberg used longitudinal data from the Youth in Transition study to explore the causal relationships between delinquency and self-esteem. The present study is based on the same sample of young men in high school and extends Rosenberg and Rosenberg's analysis, first by using the same cross-lagged correlation methods applied over a longer time period, and then by employing a â€œcausal modelingâ€ approach using the LISREL computer program. Each of the analyses was carried out using the total sample as well as two subsamples, the highest and lowest quartiles in initial self-esteem. The causal modeling analyses attempted (a) to take careful account of the actual periods referenced by the measures of delinquency and self-esteem, (b) to control socioeconomic status and ability, and (c) to extend the model to demonstrate ways in which participation in teenage social life and current educational attainment might also influence and be influenced by self-esteem. The analyses suggest that self-esteem plays little part in influencing the teenage behaviors and orientations that follow in time. Consistent with Kaplan's prediction, among young men who enter high school with low self-esteem, the effects of delinquent behavior tend primarily to be self-enhancing.