A sample of 63 subassertive adults participated in four 90-minute sessions of group assertion training. The treatment components of challenging maladaptive cognitions and learning self-instructions were examined by comparing Rational-Emotive, Self-Instructional, and Behavioral Assertion Training. A delayed-treatment control group was also included. The three training groups were not found to be significantly different at posttest on self-report and behavioral measures of assertiveness. The instruments used for self-report were the Rathus Assertive Schedule, the Social Avoidance and Distress Scale, the Fear of Negative Evaluation Scale, and the Self-Assertion Scale. Generalization of treatment gains was successfully demonstrated for all three training groups on role-played scenes not previously rehearsed. However, on an "in vivo" test of transfer of training at posttest, only the Rational-Emotive group evidenced significant generalization of treatment gains. Improvements in assertiveness were shown to be successfully maintained at three-month follow-up for all three training groups. These findings were discussed in terms of the specific techniques employed to facilitate maintenance (e.g., use of personalized assertive scenes for role playing and self-monitoring during follow-up). (Author)