A pre-post comparison group design estimated the impact of the program on ultimate outcomes of 30-day use of alcohol, tobacco, and other drugs and intermediate outcomes of risk behaviors, psychosocial well-being, and school involvement, as measured by self-administered questionnaires given each fall and spring between 1993 and 1996. A nonequivalent comparison group was drawn from nonparticipants at each school. Data analysis relied on a multivariate model comparing ultimate and intermediate outcomes at follow-up among treatment and comparison students after adjusting for demographic, baseline risk, and program exposure differences for treatment students. Results showed the follow-up rate for treatment students was 71 percent at the middle school and 48 percent at the high school. Comparison group students differed from treatment group students on selected demographic and behavioral measures at baseline. Unadjusted outcomes at follow-up were significant worse in the treatment group than in the comparison group. Statistical adjustment eliminated significant differences between treatment and comparison groups at follow-up, and high program exposure was associated with better outcomes for some measures. The authors conclude adjustment for differences between treatment and comparison groups and correction for variation in program exposure in the treatment group are essential for accurately estimating benefits of drug prevention programs. 12 references and 5 tables