Seventy-seven cocaine-using methadone patients were enrolled in a 6-month, structured, manual-driven, cognitive-behavioral treatment program. Sessions consisted of five individual or group sessions per week. At intake subjects showed extensive polydrug abuse, psychiatric comorbidity, criminal histories, and HIV risk behaviors. Treatment intensity was measured by dividing number of sessions attended into quartiles. Paired comparisons, within treatment quartiles, were made between subjects' intake and 6-month self-reports of cocaine use. Subjects in quartiles two through four showed significant reductions in frequency of cocaine use at follow-up, with subjects who received the most treatment showing the greatest reductions in cocaine use. Bivariate and multivariate analyses showed that treatment sessions attended remained a strong predictor of reduction in cocaine use at follow-up, even after controlling for drug use at intake and background variables. The results indicate that there is a substantial treatment dose-response relationship. Tables, figure, references