This study investigates the effects of a reinforcement contingency (Chinese Toys) on subtraction performance in terms of proficiency and acquisition. An eleven year old boy and a twelve year old girl participated in this study. They were labeled learning disabled and were unable to perform subtraction problems which required borrowing. 　　The experiment followed a multiple baseline design: the two subjects were given the same experimental intervention sequentially. At the same time each day, the subjects were allowed four minutes to work on a daily problem sheet: two minutes on Type A problems and two minutes on Type B problems. Type A problems were subtraction fa□□s (i.e., -814). The subjects had median score of about 30% correct for these problems during the baseline phase. Type B problems required borrowing (i.e., -824). The subjects scored 0% correct on these problems. When reinforcement was added, the subjects' scores on both types of the problems gradually rose to 100% correct. The correct rate medians for the subjects also increased markedly. When the reinforcement contingency was withdrawn, both subjects maintained their high level of performance. 　　These results indicated that reinforcement not only increased the children's computational proficiency, but also improved the children's computational acquisition and accuracy. It suggests that reinforcement is a powerful educational strategy effective in some situations if the learning objective adequately matches the children's ability.