A quantitative method for typing psychopathology of youngsters with mental retardation was assessed. Parents of 601 students in special educational placements rated them with the ABC. The sample was randomly divided into two subgroups. We used Ward's hierarchical method of cluster analysis and derived eight clusters in Subsample 1 and satisfactory agreement, kappa = 0.64, for the eight-cluster solution in Subsample 2. Clusters were: Problem Free, Within Normal Limits, Conduct Problem, Shy/Inactive, Hyperactive, Social Withdrawal With Agitation, Undifferentiated Behavior Disturbance, and Autistic-Like Behavior. We also used demographic (age, gender, IQ, physical handicaps) and clinical variables (DSM diagnoses and target behaviors) extracted from school records of 228 participants to appraise validity of the empirically derived clusters. The external data provided modest to strong support for all except one cluster (Social Withdrawal With Agitation).