Are genetic and environmental risks for adolescent substance use specific to individual substances or general across substance classes? We examined this question in 645 monozygotic twin pairs, 702 dizygotic twin pairs, 429 biological sibling pairs, and 96 adoptive (biologically unrelated) sibling pairs ascertained from community-based samples, and ranging in age from 12 to 18 years. Substance use patterns and symptoms were assessed using structured psychiatric interviews. Biometrical model fitting was carried out using age- and sex-specific thresholds for (a) repeated use and (b) problem use, defined as one or more DSM-IV symptoms of abuse or dependence. We hypothesized that problem use would be more heritable than use in adolescence, and that both genetic and environmental risks underlying tobacco, alcohol, and marijuana use and problem use would be significantly correlated. Results of univariate analyses suggested significant heritable factors for use and problem use for all substances with the exception of alcohol use. Shared environmental factors were important in all cases and special twin environmental factors were significant for tobacco use, tobacco problem use, and alcohol use. Multivariate analyses yielded significant genetic correlations between each of the substances (for both levels studied), and significant shared environmental correlations among use variables only. Our results suggest that tobacco, alcohol, and marijuana problem use are mediated by common genetic influences, but shared environmental influences may be more substance-specific for problem use.