Children in a first-grade dassroom were divided into four groups. Baseline measures of disruptive classroom behavior were taken on a well-behaved and disruptive child in each group. Following baseline, four types of token economies were simultaneously introduced and rotated every 10 days within a Latin Square design. The token economies were: (1) individual reinforcement determined by individual performance; (2) group reinforcement determined by the behavior of the most disruptive child; (3) group reinforcement determined by the behavior of the least disruptive child; (4) group reinforcement determined by the behavior of a randomly chosen child. The token economies were compared on their effectiveness in changing target behavior, preference by the targets, ease of use, and cost. Additionally, sociometric responses were taken on questions of responsibility, friendship, and funniness. Results showed a significant decrease of inappropriate behavior for the disruptive children and no difference between the effectiveness of the four types of token economies in producing behavior change. However, there were other differences that indicated that the system in which group reinforcement was determined by a randomly selected child would be desirable for most teachers. Results also showed changes in the sociometric status of the disruptive children. As predicted, disruptive children were rated as more responsible when they were in the group reinforcement determined by the most disruptive child in the group token economy. Using behavior modification techniques indirectly to change sociometric status is suggested as offering a new potential technique for behavior change agents.