Data focused on illegal alcohol and marijuana use were collected in 1981 from 4,859 high school students living in two rural counties in southern Georgia. The sample consisted of over 85% of all students in grades 8 through 12 within the two counties. The data were used to test a facilitative-constraint model developed from selected components of differential association, differential indentification, and social control theories. Discriminant analysis was used to compare "users,""experimenters," and "non-users" on the basis of factors chosen to represent each dimension of the theory. The findings demonstrated that the theoretical model was useful in correctly classifying students into appropriate categories for both drugs assessed, but was particularly useful for predicting marijuana use. The most important empirical finding was identification of two variables which were demonstrated to be good discriminators of drug use: "access to drugs" and "identification with pot-head groups." Facilitative indicators were the best in predictive variables, while constraint factors were shown to be much less important in differentiating among the various drug user groups. Programs designed to increase identification with conventional groups while simultaneously reducing identification with drug-oriented groups will probably be effective in reducing illegal drug use. (Author/BRR)