The usefulness of an approach to reduce impulsive behavior in 11 institutionalized behavior disordered boys (10 to 16 years old) was investigated. Ss were taught to use five steps to solve cognitive problems. Steps involved task analysis, were modeled by the experimenter followed by reinforcement of verbalization of the rules, and included active controlled responding by the Ss. Twelve control Ss were exposed to the same tasks, encouraged to do as well as possible, and were reinforced for correct responding in the same manner as the treatment group. Pre- and posttest results on the Porteus Maze test Q score and on the latency score of the Matching Familiar Figures (MFF) test revealed significant differences. Q scores decreased for the experimental Ss but stayed the same for controls. Both groups demonstrated an increase in MFF latency scores. (CL)