This paper presents the rationale for a 24-week, randomized trial designed to test whether risperidone plus structured parent training would be superior to risperidone only on measures of noncompliance, irritability and adaptive functioning. In this model, medication reduces tantrums, aggression and self-injury; parent training promotes improvement in noncompliance and adaptive functioning. Thus, medication and parent training target related, but separate, outcomes. At week 24, the medication was gradually withdrawn to determine whether subjects in the combined treatment group could be managed on a lower dose or off medication without relapse. Both symptom reduction and functional improvement are important clinical treatment targets. Thus, experimental evidence on the beneficial effects of combining pharmacotherapy and exportable behavioral interventions is needed to guide clinical practice.