From 1976 to 1980, 377 randomized patients were entered in a multicentric study supported by the W.H.O. Melanoma Group. In this report, only 196 patients are considered for analysis: 31.6% were excluded because of inadequate treatment or protocol violations. Three arms of treatment were compared: DTIC alone, DTIC plus bacillus Calmette-Guérin, and DTIC plus Corynebacterium parvum. No significant difference in terms of complete response plus partial response was detected among the three regimens. Twenty-five patients (12.7%) responded with complete tumor regression, and 18 (9.3%) with partial regression. Complete plus partial responses were thus achieved in 22% of evaluable patients. Responders had a longer median survival than nonresponders. Overall median survival was 6 months. Brain metastases were detected in 17% of the patients. Median time from the beginning of treatment to the diagnosis of CNS involvement was 4 months. There is no evidence that active aspecific immunotherapy added to DTIC increases the response rate in patients with disseminated disease. The toxic effects were acceptable, and no drug-related death was observed.