Two experiments using paraprofessional personnel were conducted to determine effectiveness of a schoolwide token system on academic performance of trainable mentally retarded (TMR) children. The first experiment consisted of 42 TMRs in a 6-week summer program, in which reading, writing, and arithmetic were the behaviors studied. Pretest and posttest measures on the Metropolitan Readiness Tests indicated that childrens' scores increased over 25% following only 18 hours of token reinforcement for academic performance (p greater than .001). The second experiment was designed to isolate the relative role of reinforcement on the childrens' academic improvement. The experiment involved 31 chidlren who were able to choose a class with or without reinforcement in writing and arithmetic. Results showed that TMRs chose to work on the subject area having reinforcement and also, that the childrens' correct academic performance increased significantly as a function of reinforcement. It was concluded that TMRs could profit from academic training when an extrinsic motivational system was involved. (CB)