The purpose of this study is to examine how the assertive-aggresive behavior is maintained, increases and changes its response and victim in the interactions of nursery school children. It is predicted that the consequences of the assertive-aggresive behavior supplied by the peer group work as positive or negative reinforcers and condition the assertive-aggressive behavior. In order to confirm this prediction, the two following hypotheses are tested. 1. When an assertive behavior is followed by a positive reinforcer, on the next occasion in which the same aggressor is involved he would select the same aggressive response and the same victim. On the other hand when his behavior is follwed by a negative reinforcer, he would change either the aggressive response or the victimor both. 2. If a child who initially displayed few aggressive behaviors is frequently victimized and counterattacks successfully, his aggressive initiations would incerease. The subjects were 4o children of three groups in two nursery schools. Their ages in each group were 1:6-2:5 (mean 1:10), 1:6-2:4 (1:11) and 2:8-3:5 (3:1). Their aggressive interactions were observed for 11 weeks. The aggressive responses were classified into 4 categories, the responses of victims into 7 categories and the responses of teachers into 4 categories. The responses of victims and teachers were labeled a priori as positive or negative reinforcers according to Patterson et al.(1967) and the pirot study. A total of 2477 aggressive behaviors weve observed. Hypothesis I was confirmed by two groups, but not by one group aged 1:6-2:5. So the patterns of aggressive interaction of the three groups were analyzed. The result suggested that the amount of the peer interaction or the responsiveness to the stimuli supplied by the peer group would decide the conditionability of the aggressive behavior. As for the hypothesis 2, 11 children who had initially displayed few aggressive behaviors were followed up, and the results all satisfied the prediction.