Children evidencing conduct disorders comprise the bulk of clinical referrals. Longitudinal studies have found that the prognosis for these children is poor in that the majority exhibit antisocial behavior in adulthood. In order to compare two methods of treatment, relationship-based and contingency management, 53 children (33 boys, 20 girls), aged 4-7, who had been referred to a community mental health clinic for conduct disorders were assigned either to relationship-based treatment, contingency management, or to a waiting list. Pretreatment assessment included parental interview, personality and IQ testing, and home observations. Posttreatment assessment included home observations and feedback measures. Throughout the study the Dyadic Parent Child Interaction Coding System (DPICS) was used to monitor child deviance and noncompliance. In both treatment conditions, families were seen individually for seven or more videotaped sessions. Relationship-based training consisted of teaching the parents interaction skills in regard to play, prosocial behavior and social reinforcement. Contingency management training consisted of identifying and changing deviant behavior through point charts, homework assignments, and rewards. According to parent report measures both treatment groups improved relative to the control group. However, home observations revealed that only children assigned to the relationship-based approach significantly decreased deviance and noncompliance, which may be due to an emphasis on coaching parents in the use of social reinforcement. (BL)