This series of studies created and validated a role-play measure of cigarette, alcohol, and drug refusal skills, the Behavioral Substance Refusal Test (BSRT) for seventh and eighth graders, which included a measure of self-efficacy and outcome expectancies, the Role-play Self-Efficacy and Outcome for Refusal Test (RSORT). A series of focus groups asked seventh and eighth graders to discuss specific situations involving peer pressure to use drugs. These discussions were used to create the scenes for the behavioral drug refusal measure, which was based on the Behavioral Assertiveness Test for Children - Revised (Ollendick, 1981) and included both verbal and nonverbal components of aggression. Two additional categories of assertive behavior, statement of consequences and verbal repertoire, inductively determined from previous research on behavioral measures of assertion (Weist & Ollendick, 1991 ), were added to enhance the measure. In the first study, a multi-trait multi-method matrix (MTMMM) examined the measure's internal consistency and construct validity. The BSRT demonstrated promising convergent validity, but it did show a large amount of method variance. This method variance indicated that role-play and self-report measures of drug refusal skills measure different constructs. Further studies must determine whether the behavioral or self-report measures of assertion are more valid. The MTMMM indicated that the relationships between the constructs assertion, aggression, and passiveness may be more complex than is indicated in previous research on assertion, and suggestions are made for future measures of these constructs. In the second study, peers rated the seventh and eighth graders' role-play performances and confirmed that subjects who received high drug refusal scores on the BSRT were not rejected by peers and were perceived as effective by their peers.