The effects of four reinforcement conditions (individual consequence, group-shared consequence, and two different proportions of individual and group-shared consequence) on (a) peer tutoring, (b) arithmetic performance, (c) studying, (d) nonstudying, and (e) disruptive behavior were measured employing 60 experimental and 34 comparison children from three fifth-grade classes in an inner-city school. The 100% shared consequence consistently produced the highest incidence of peer tutoring as well as the greatcst increment in the number of correct problems, both within and between periods. As the proportion of shared consequences decreased, the number of problems worked correctly decreased. Concurrently, the incidence in peer tutoring decreased as the proportion of shared consequences decreased. All consequences maintained high rates of studying behaviors and low rates of nonstudying and disruptive behaviors. The results of this study suggest that shared consequences may be useful in creating cooperative work patterns and increasing arithmetic performance in classroom settings.