Project Alert, the curriculum evaluated, is designed to motivate students not to use drugs and to teach them the skills required to translate this motivation into effective resistance behavior. So as to overcome weaknesses in previous evaluations of similar programs, the program was evaluated in a variety of environments. The evaluation achieved well-balanced experimental groups, and the program was implemented as designed. Reliable outcome measures were obtained, and alternative explanations for the results were eliminated. In most cases, multiple strategies were used. Extensive analyses assessed how well the strategies worked; they indicated that each obstacle was overcome. The evaluation results show that Project ALERT prevented or reduced both cigarette and marijuana use during the junior high years. It was particularly successful in delaying the onset of marijuana use and in holding down occasional and regular smoking among previous cigarette experimenters. Of particular importance in the evaluation design was the recruitment of institutions from diverse communities and the maintenance of their cooperation over several years. Also important were the techniques for facilitating faithful program delivery and the monitoring of the implementation process. Recommended design and analysis features include the use of more than simple random assignment to achieve a balanced design and the use of control variables to rule out alternative explanations of the results. 6 tables, 16 notes, and 37 references