Data were collected on a sample of 3,927 children in kindergarten through fifth-grade. A sample of children (n=80) in grades three through five were identified as officially delinquent. This delinquent sample was matched with 10 control groups from the same original cohort (n=827). The sample predominantly consisted of ethnic minorities (70 percent). The study found that both child self-reports and teacher ratings of aggression and prosocial behaviors discriminated between officially delinquent versus nondelinquent children. Also, both methods correctly classified about two-thirds of delinquent and nondelinquent children. Further, child self-reports and teacher ratings of female delinquency status improved classification over chance by approximately 45 percent; the classification of male official delinquency status improved by 17 percent and 35 percent for child self-reports and teach ratings respectively. In summary, teacher ratings were superior classifiers of official delinquency status in elementary school children. This was particularly true for males, less so for females. Implications for the classification of delinquent behaviors in young children are discussed. 4 tables, 2 figures, and 38 references