Nonassertive 5s received two sessions of training in refusing unreasonable requests. The training was a standardized, semiautomated, laboratory analog of behavior rehearsal therapy, consisting of three main components: overt or covert response practice, symbolic verbal modeling, and therapist coaching. Compared to control Ss, who were led to believe that the assessment procedures were a form of behavioral training, the behavior rehearsal 5s improved dramatically in their assertive-refusal behavior on self-report and behavioral laboratory measures. Covert rehearsal tended to produce the greatest improvement. The experimental methodology employed in this study proved to be a promising approach to the evaluation and development of behavior rehearsal therapy.1 Copies of the assessment and training stimuli used in this study are available upon written request.