This paper is a preliminary report of the construction and partial validation of a nonverbal test designed to assess aggressive behavior in chldren. The test is of the multiple choice variety, making use of cartoons for both the stimulus patterns and the response alternatives. The test identifies 2 sets of variables deemed necessary for the making of predictions about aggressive behavior in children. The ariables represent a translation of the frustration-aggression model. The data indicate a significant age difference operating in the kinds of response to frustration chocen. There is a consistent and significant increment to the prediction of concurrent aggressive behavior when combinations of response dispositions and inhibitory varibles are used instead of using aggressive response variables alone. Using both types of variables also results in an appreciable drop in teh amount of false positive error as compared to using aggressive response variables alone.