The primary data source for subtyping subjects and for coding validating variables was an offender's extensive clinical file which included all information obtained during the patient's evaluation and commitment periods at the Massachusetts Treatment Center. Postcommitment information included treatment summaries, behavioral observation reports, work reports, summaries of program participation, and results of any diagnostic assessments. Data also covered past institutionalization records, school and employment reports, police reports, court testimony, parole summaries, probation records, and social service notes. Although study results supported one a priori hypothesis, they disconfirmed two others and yielded an unpredicted but theoretically interesting abductor covariate. The hypothesis that child abductors would more likely be classified as "low" in their contact with children (have little or no contact with children outside of their offenses) than the nonabductors was supported. In contrast, the hypotheses that the abductors were more likely to be characterized by a history of antisocial and criminal behavior as well as a greater degree of aggression were not supported. Abductors were found to be lower in social competence than nonabductors. Findings suggest that abduction as a victim control strategy is more likely to be used by offenders with poor social and interpersonal skills. The complex interrelation among social competence, weapons, and sadism for abductors and nonabductors is explored. 6 tables and 10 references