This study examined the comparability and predictive validity of three approaches to the measurement of child maltreatment. Adolescents (N = 160, aged 11-17) were randomly selected from the open caseload of a child protection agency. Global ratings of maltreatment severity were made by three reporting sources: researchers on the basis of protection agency case files, protection agency social workers, and the adolescents themselves. Ratings were made of five types of maltreatment: physical, sexual, emotional, neglect, and exposure to family violence. Self-reported (YSR) and caretaker-reported (CBCL) adjustment measures were also obtained for each subject. Results indicated that over 90% of the sample had experienced more than one type of maltreatment. Comparison of ratings across sources indicated considerable disagreement with respect to judgments of maltreatment occurrence and severity. Relative to professional ratings, adolescent ratings were better predictors of externalizing and internalizing symptomatology in both univariate and multivariate analyses.