Past studies have shown that disruptive behavior can be eliminated and attending behavior accelerated in an academic setting. The relationship between these behaviors and academic performance is not well understood. The effects of behavioral and performance contingencies on classroom behavior and on academic performance were investigated. The subjects, third‐grade students from an inner city elementary school, were exposed to a series of conditions including baseline, behavior contingencies, performance contingencies, and a mix of behavior and performance contingencies using a reversal design. The students worked 100 randomly selected mathematics problems for 20 minutes each day during each period. Behavioral contingencies improved attending and decreased disruptions but did not improve performance. Performance contingencies increased per cent correct problems but attending declined and disruptions increased. The combined contingencies increased both performance and attending. The experiment was replicated with another class of children varying the sequence of conditions and the amount of token reinforcement that could be earned. The findings emphasized the importance of designing specific contingencies for specific target behaviors. Behavioral contingencies did not have the positive effect on performance often implied, nor were performance contingencies alone able to maintain acceptable classroom behavio