Literature on child abuse supports the conception that physical abuse is a multidetermined behavior. Multifaceted treatment programs have shown some promise in dealing with the heterogeneity of abusive parents. Most of the reported comprehensive intervention programs have provided predetermined doses of a variety of treatment components to each subject. The intent of this study was two-fold: (a) to provide treatment components based on assessment of the parent and (b) to train the parent to a specified level of competency. A multiple baseline design was used in this clinical study of six agency-referred, physically abusive parents. Based on initial assessment and ongoing observation, subject parents were provided with one or two of four available parent-training components (child behavior management, cognitive modification, relaxation, and systematic desensitization). Treatment effects on 16 dependent variables were measured by self-report, coded audiotape, coded observation, physiological measures, and reports of abuse to public agencies. Results indicated improvement by all the subjects on most of the dependent variables (i.e., 77 of 90 comparisons). However, only three of the six subjects met all of the predetermined criteria for termination of all intervention. Subjects met 15 of 22 training competency criteria. Reductions in abuse indicators were maintained on most of the dependent measures during 30- and 90-day follow-up probes. Only one subject was re-reported for child abuse during the year following treatment. The low attrition rate was seen as a function of assessment.