The hypothesis that risk for substance abuse is not unitary was tested by a log linear regression analysis. A series of risk factors was evaluated in a sample of 219 nonhospitalized, nontreated, young male and female subjects. Risk for alcohol abuse/dependence was independent of risk for drug abuse/dependence and the risks were additive in the prediction of lifetime substance abuse. Antisocial personality diagnosis predicted alcohol abuse/dependence, family history of drug abuse predicted drug abuse/dependence, and, additively, they predicted alcohol and drug abuse/dependence diagnosis over a lifetime. These data support a hypothesis that risk for substance abuse is not unitary.