The Drug Use Survey was administered via a direct mail to a simple random sample of 2200 students enrolled at a large southwestern U.S. university. A purpose of the study was to determine the predictability of self-reported drug use utilizing motivations (personal and social) commonly reported by substance users. Two subscales were developed, one for each category of motivations. Reliability for each subscale as estimated by coefficient alpha was .80 and .86, respectively. A series of step-wise multiple regression analyses were computed in which individual drug use indices served as criterion variables, while the predictor variables were the personal motivations subscale (PMS) and the social motivation subscale (SMS) for each model. The results indicated that the PMS was the stronger predictor in every model with the exception of the model that predicted the alcohol use index. The SMS was the best predictor for alcohol use.