We compared the predictive validity of five aggregation methods for multiple informant data on child and adolescent behavior problems. In addition, we compared the predictive validity of these aggregation methods with single informant scores. Data were derived from the Minnesota Longitudinal Study of Parents and Children (N = 175). Maternal and teacher reports of child behavior problems were obtained at ages 7 and 16 using the Child Behavior Checklist (CBCL) and Teacher Report Form (TRF). Adolescent self-report of child behavior problems was obtained at age 16 (Youth Self-Report). Adolescent psychopathology data was obtained at age 17.5 using a semi-structured diagnostic interview (The Kiddie Schedule for Affective Disorders and Schizophrenia; K-SADS). Results generally showed that the latent variable performed better in predicting adolescent psychopathology compared to the other aggregation methods and single informant scores. The other aggregation methods did not, however, outperform the single informant scores. (Contains 4 tables.)