A multiple-baseline design of single-case experiment and a multiple time series design of quasi-experiment were combined to study the effectiveness of home-based reinforcement on the academic performance. Two classes of fifth grade students were assigned to three groups according to the willingness of their parent to participate the home-based reinforcement program. The home-based reinforcement in the Experimental Group was manipulated by the experimenter, whereas it was not in the Control Group I; Those, whoses parents could not participate the home-based reinforcement program, constituted the Control Group II. The dependent variables were quizzes and monthly exams in Reading and Arithmetic. Scores were transformed into T scores to eliminate the curve fluctuation caused by the different difficulty of each test. Results showed that the academic performance of the Experimental Group and the Control Group I were better than that of the Control Group II. No difference was found between the Experimental Group and Control Group I. It also supported the notion: "If token programs serve as a priming or incentive function, one would certainly expect academic behaviors to be more difficult to chang than social behavior."