The analysis revealed that nearly 38 percent of those who tried marijuana continued using it. Potential predictor measures were grouped into a drug-specific category and a social context category. Logistic regression models were used to determine the effects of each set of factors on continued use, controlling for background characteristics. Results revealed that only the drug- specific factors had a statistically significant effect on the likelihood of continued use. Students who felt that the adverse physical and psychological effects of marijuana were not very important reasons for discontinuing use and those who had become stoned during their experimental stage of use were the most likely to continue use after initiation. Findings suggested that the perceived physical and psychological effects of marijuana are more important determinants of continued use than are social factors or benefits related to use; any relationships between social factors and continued use are mediated by the perceived effects and risks of the drug. Tables, note, and 31 references (Author abstract modified)