Using data from the National Education Longitudinal Survey of 1988, the authors assess the extent to which adolescents' levels of parental attachment and opportunities for participating in delinquent activities mediate the family structure--substance use relationship. A series of hierarchical regressions supported the hypotheses that high levels of substance use among adolescents residing with stepfamilies would be explained by low parental attachment, whereas heightened opportunities for participating in deviant activities would account for the substance use behaviors of individuals living in single-parent households. More generally, the findings suggest that family structure has a moderate effect on youth substance use; that parental and peer relations are better predictors than family structure of levels of alcohol and marijuana consumption; and that variations in parental attachment, parenting style, and peer relations across family types explain some, but not all, of the effects of family structure on adolescents' substance use behaviors. (Contains 4 tables, 4 figures, and 5 notes.)