Several hundred children were involved in initial data collection, but the author concentrated on the outcome for children involved in delinquency. Data on social, educational, and individual differences were analyzed to establish factors that would predict later criminality and offense type, as well as responsiveness to therapeutic intervention. It was found that a research-based approach to dealing with behavior problems and delinquency is viable as long as it incorporates the school context. Certain principles need to be applied if the process by which an individual achieves or fails to achieve perceived objectives is to be made explicit. A framework of formal analysis is necessary to understand persistent criminality in children. Understanding, however, is not enough; it needs to be followed by agreement on objectives and intervention. When working with children, school programs and objectives must be periodically evaluated and changed. 69 references and 1 figure