Twenty-seven male and female delinquents who had either recently been placed out of the home or for whom placement was imminent were court referred to in-home, time-unlimited family therapy (mean of 16 sessions). A comparison group of 27 lower risk delinquents received only probation. The intervention was divided into three phases: assessment, therapy, and education. The assessment phase consisted of the appraisal, by observation and interview, of family behavioral patterns and the reinforcers maintaining those patterns. The therapy phase was designed to alter attitudes, expectations, cognitive sets, labels, emotional reactions, and perceptions of relationships between family members so as to reduce blaming and to portray each member as an unwitting victim of a poor learning history. The education phase was designed to either teach family members skills they lacked or prompt them to use skills they already possessed. Outcome was measured by the number and severity of offenses during 2 1/2 years following group assignment. The delinquents who received the family therapy had an 11-percent recidivism rate compared to 67 percent for the comparison group. Sex differences are noted, as well as differences between Alexander's studies and this study, which may account for the improved outcomes with more dysfunctional families. 2 tables and 32 references (Author abstract modified)