Nine seventh- and eighth-grade students who had a history of poor math achievement were selected to participate in the study. Base rate data was gathered which indicated the percent of time the students were engaged in behaviors relevant to the assigned academic tasks. During treatment a clock buzzer apparatus was placed in the room, and the students were told that they could earn free-time contingent on concomitant task relevant behavior by all the class members. The clock was allowed to run when the entire class was working on the assigned materials, which provided a visual display of the cumulative earned free-time. When any of the students behaved inappropriately, the clock was turned off and a buzzer was sounded. The group contingent conditions substantially increased the percent of task relevant behavior emitted by the subjects. A treatment reversal reduced the level of appropriate behavior, and reintroduction of group contingencies again markedly increased the group level of task relevant behavior. Alteration of teacher contingencies had little apparent effect on the students' behavior. (Author)