Self-evaluation training was operationalized as a training package comprised of the following components: systematic contingent reinforcement intended to promote acceptable rates of appropriate behavior prior to the introduction of self-evaluation training procedures; self-evaluation, and procedures designed to promote accurate self-evaluation; tokens and back-up reinforcement; and teacher praise. Training was provided to a group of three seriously emotionally disturbed students, ages 9, 10, and 11, who received their education in an institutional setting. Then, in a multiple baseline design, subjects self-evaluated in their classroom to determine if generalization of high rates of appropriate behavior demonstrated in the training setting would occur. Data indicate that generalization to the classroom could be claimed for each of the three subjects.