The data gathered at two California Youth Authority institutions for boys. The program at one school was based on transactional analysis, and the other was grounded on the principles of behavior modification. The subjects included all boys from 15 to 17 years of age assigned to the institutions from August 1969 through March 1971. A total of 823 subjects included whites (56 percent), blacks, and Mexican Americans. All boys were placed in one of the nine personality categories by use of the Sequential I-level Classification System, a semiobjective approach to classification using data obtained by such means as an interview and a sentence completion test. Delinquents' behavioral change while in the treatment program was measured by a questionnaire containing 80 behavioral items for which the rates gave an estimate of frequency of occurrence. Factor analysis was used to define 14 scales. Reconviction data obtained in 1975 from the California Youth Authority's data information system were used to determine parole outcome 6 months after release. An analysis of variance was used to examine the difference between the change in pretest and posttest scores for recidivists and nonrecidivists on each of the 14 scales. Results showed an interaction between recidivism status and personality characteristics on change scores for several of the scales. For instance, the nonrecidivists who were labeled Neurotic Acting-Out and Immature Conformist improved on several scales, while those classified as Neurotic Anxious who did well on parole showed less improvement while institutionalized than those who later recidivated. Reseachers should use several types of scales to measure change when analyzing results. Various treatment interventions and different treatment goals may necessary for different groups of delinquents. For example, for some groups the most appropriate type of therapy might by some form of behavior modification and/or social skills training designed to stimulate change along the relevant behavioral dimensions, while for others a more internally oriented, psychodynamic therapy which encourages expression of anger might be better. Discussion of prior studies, statistical data, footnotes, and 21 references are included.