Drug abuse among adolescents and young adults is a serious social problem and a public policy concern in the United States. Previous research has well-documented the prevalence of illicit drugs on college campuses. The goal of the current study was to add to the literature in this field by examining how illicit drug use in college is influenced by demographic variables, family and background characteristics, peers, experiences of alcohol and tobacco use, and academic activities. Participants were a sample of 1,218 college students attending 12 post-secondary institutions in 8 southern States. Participants completed a self-administered survey during the fall 1998 academic term. The 56-item survey instrument measured a wide range of individual characteristics, academic activities, family characteristics, and peer influences and activities. Two dependent variables measured drug use. Separate analyses were conducted to construct profiles of individuals who use marijuana only as opposed to individuals who use harder drugs, such as cocaine, stimulants, or LSD. Results of logistic regression analysis revealed that students who use marijuana only are more likely to have received little or inconsistent supervision as children, are members of fewer clubs and organizations, are more likely to skip class, smoke, party with friends, get drunk often, and get drunk in public than students who do not use marijuana. Students who use harder drugs are more likely to have received little or inconsistent supervision as children, get drunk frequently and in public, be behind others in school, smoke tobacco, and spend their free time at friends’ houses or at bars where they are regulars. This analysis has built on previous research by assessing a wider range of variables than has previously been examined. Two measures in particular emerged as significant indicators of marijuana use only and hard drug use: one’s mother being employed full time and one’s parents usually knowing where they are. Practical applications of this study include the development of programming designed to specifically address the variables identified as predictors of drug use. Tables, references