The task of this study was to determine whether the twelve subtests of the Student Behavior Description Questionnaire (SBDQ) could distinguish behavioral problem from nonbehavioral problem students. The population consisted of all seventh and eighth grade students enrolled in five intermediate schools in a large suburban school system in Virginia. The sample consisted of two groups from each school. The first group was composed of students who have been referred for demonstrating negative behavior. The second group was selected from all other students who had not had discipline problems. The sample included 67 behavioral problem and 67 non-behavioral problem students. Stepwise discriminant analysis was used to determine the value of the SBDQ subtests in distinguishing behavioral problem from nonbehavioral problem students. Using the discriminant function prediction equation, the SBDQ was able to predict group membership (behavioral problem vs. nonbehavioral problem) with the use of the five subscales: Academic, Disinterest, Friendship, Participation, and Inclusion. The two measures used to assess the degree of congruence between predicted and actual classification of students were (1) the precentage of correct classifications and (2) tau, a measure of the reduction of error in group classification resulting from knowledge of a set of independent variables. In this study, the discriminant function correctly classified 76.1% of the students and tau indicated that use of the discriminant function in predicting whether a student would be a behavioral problem or not would reduce the number of errors in correct classification by 52%. / Ph. D.