The purpose of this dissertation is to contribute to understanding the children`s mechanisms of aggressive behavior revealing that aggressive and non-aggressive children, for what reasons or through what processes, become more aggressive or less aggressive. The hypotheses are as follows; First, under the frustrative situation in which the intention is ambiguous, aggressive children will show aggressive behavior than non-aggressive children. Second, under the frustrative situation in which the intention is ambiguous, aggressive children will be weaker in the tendency to delay showing aggressive behavior than non-aggressive children. The subjects consisted of 60 boys (aggressive children ; 30, non-aggressive children ; 30) Who were selected out, through the rating by each teacher in charge and nomination from the peers, among 496 children who were in the second grade of an elementary school located in middle-low class. The data of each child collected at the experiment were verified by the x^2 and t test for Hypothesis 1 and Hypothesis 2. The consequences found in this experiment are as follows ; first, aggressive children in comparison with non-aggressive children in aggressive behavior showed non-verbal direct aggressive behavior, non-verbal indirect aggressive behavior, and verbal aggressive behavior more by a significant difference and supported Hypothesis 1. Second, not only were there more aggressive children who could not delay the expression of aggressive behavior than non-aggressive children, but there was more tendency not to delay the expression as the delay time was short. Thus Hypothesis 2 was supported as well.