Describes pilot findings from a treatment development study aimed at improving treatment for comorbid depressed and chemicallydependent patients. A comparison of standard RCT analyses with Hierarchical Multiple Regression (HLM) procedures revealed the latter to be more sensitive to the relative effects of patient, treatment, and patient-treatment matching variables among a small sample of (N = 40) depressed, stimulant-abusing patients. Participants were randomly assigned to one of three treatments, a standard Cognitive Therapy for Drug Abuse (CT), a contrasting Cognitive-Narrative Therapy (NT), and a Prescriptive Therapy (PT), the latter of which selectively applied procedures from both of the other two treatments following an Aptitude × Treatment Interaction (ATI) model. The results supported a multiple factor view of psychotherapy effects, including the hypothesis that patient, treatment, relationship, and patient-therapy matching variables add independent power to the prediction of treatment outcome, especially during follow-up. ATI effect sizes were stronger than those associated with specific treatments.