Of these, 1976 and 1983 followup data were available for 947 subjects. Subjects were classified as terminators, occasional offenders, and persistent offenders on the basis of subsequent records. Presentence investigation reports were examined to test 12 hypotheses drawn from the literature dealing with the following factors: family and development, social deviance, socioeconomic status, educational and vocational adjustment, interpersonal relations, adjustment, and criminal behavior. Analyses of variance provide support for eight hypotheses. The three offense groups differed with respect to age at first arrest, length and seriousness of juvenile conviction record, and the extent to which they had problems at work, with other people, and with authorities. They also differed in achievement motivation and personality test scores. In all cases, persistent offenders had the most deviant and least socially desirable scores, while eventual terminators had the most positive characteristics. Despite these differences, their magnitude was not sufficiently great to enable accurate prediction of subjects' later criminal behavior. Implications of these findings for specialized treatment programs and selective incapacitation are discussed; 58 references are provided.