The purpose of this study was to describe a sample of incarcerated juveniles with severe delinquency histories using variables selected from the MMPI-A, WISC-III, DVMI, WRAT-III, WJ-R, CPT, and WCST. The subjects were classified into subgroups using cluster analysis. The membership and profiles of the subgroups that were derived using MMPI variables were compared to the subgroups formed using variables from all domains and to previous school based diagnoses. Group membership did not correspond to school based diagnoses or with clusters formed with different sets of variables. The MMPI-A profiles were more interpretable if the clusters were formed with only MMPI-A variables.For the entire sample, these subjects were found to be about one standard deviation below the mean in Full Scale IQ, verbal expression, visual-motor integration, and academic achievement. They were near the mean of normative samples for receptive verbal language and ability to inhibit perseverative responses. Phonological processing was better than sight word reading, but not as good as reading comprehension. Executive functioning data supported the underactive cortex theory of psychopathy and favored Eysenck's (1977) hypothesis that this interferes with conditioning, but not Hare's (1970) inference of perseverative response patterns.Whether formed from variables from all domains or from MMPI-A domains only, the derived clusters resulted in a strong levels of performance effect with the best and worst performing groups in one domain functioning at a very similar level in the other domains. Groups in the moderate range were differentiated by a less well defined patternsof scores effect. / Department of Educational Psychology