Students (N = 4,466) attending 56 schools in New York State were involved in a 3-year study testing the effectiveness of a cognitive-behavioral approach to substance abuse prevention. In a randomized block design, schools were assigned to receive (a) the prevention program with formal provider training and implementation feedback, (b) the prevention program with videotaped provider training and no feedback, or (c) no treatment. After pretest equivalence and comparability of conditions with respect to attrition were established, students who received at least 60% of the prevention program (N = 3,684) were included in analyses of program effectiveness. Significant prevention effects were found for cigarette smoking, marijuana use, and immoderate alcohol use. Prevention effects were also found for normative expectations and knowledge concerning substance use, interpersonal skills, and communication skills.