PROGRAM DEVELOPMENT AND EVALUATION INVOLVE SEQUENTIAL STEPS. THESE STEPS CONCERN PROCESS IDENTIFICATION, PROCESS MODIFICATION, IMPACT OR OUTCOME, AND LONG-RANGE EFFECTS. THE GOAL OF THE IDENTIFICATION STUDY WITH DELINQUENT FAMILIES WAS TO DEVELOP POTENTIALLY MODIFIABLE MEASURES THAT WOULD DIRECTLY REFLECT THE NATURE OF RELATIONSHIPS, NOTE SYSTEM PROCESSES THAT CHARACTERIZE DELINQUENT BEHAVIOR, AND DEMONSTRATE THAT THESE PROCESSES ARE NOT EXISTENT IN ADAPTIVE NONDELINQUENT FAMILIES. A DESCRIPTIVE STUDY OF 20 DELINQUENT AND 22 NONDELINQUENT FAMILIES WAS PERFORMED USING DEFENSIVE AND SUPPORTIVE COMMUNICATIONS AS DEPENDENT MEASURES. FAMILIES WERE VIDEOTAPE RECORDED IN THE COURSE OF A 10-MINUTE DISCUSSION TASK AND A 15-MINUTE RESOLUTION OF DIFFERENCES TASK. DATA CLEARLY DIFFERENTIATED BETWEEN ADAPTIVE AND DELINQUENT FAMILIES. DELINQUENT FAMILIES DEMONSTRATED SIGNIFICANTLY HIGHER RATES ON ALL SIX CATEGORIES OF DEFENSIVENESS AND SIGNIFICANTLY LOWER RATES ON TWO OF SIX SUPPORTIVENESS CATEGORIES. NORMAL FAMILIES RECIPROCATED SUPPORTIVENESS, BUT NOT DEFENSIVENESS, BETWEEN PARENTS AND CHILDREN. DATA FOR DELINQUENT FAMILIES, ON THE OTHER HAND, REFLECTED RECIPROCAL DEFENSIVENESS BUT NOT SUPPORTIVENESS. DIFFERENCES IMPLIED THAT AN INTERVENTION PROGRAM SHOULD BE AIMED AT CONTENT (DECREASING DEFENSIVE COMMUNICATIONS), AND PROCESS (INCREASING CONTINGENT RECIPROCITY AROUND SUPPORTIVENESS AND DECREASING RECIPROCITY OF DEFENSIVENESS). INFORMATION FROM PROCESS RESEARCH GENERATED SUPPORT FOR AN INTERACTION APPROACH TO DELINQUENCY AND IDENTIFIED CLEAR AND POTENTIALLY MODIFIABLE MALADAPTIVE FAMILY BEHAVIORS. A REFINED PROCESS STUDY BASED ON OUTCOME MEASURES WAS THEN CONDUCTED TO EVALUATE CHANGES IN FAMILY PROCESSES. A 3-YEAR FOLLOWUP STUDY WAS ALSO PERFORMED, INVOLVING THE USE OF RECIDIVISM DATA FROM COMPUTERIZED JUVENILE RECORDS TO EVALUATE LONG-TERM EFFECTS OF PRIMARY PREVENTION. CONSISTENT WITH PROCESS DATA, FOLLOWUP DATA DEMONSTRATED THE CLEAR SUPERIORITY OF SHORT-TERM BEHAVIORAL FAMILY SYSTEMS INTERVENTION APPROACH OVER NO-TREATMENT CONTROLS AND THE CLIENT-CENTERED APPROACH. REFERENCES AND SUPPORTING DATA ARE INCLUDED. (DEP)