Fifty patients with metastatic malignant melanoma were randomized to treatment with either DTIC (2 mg/kg/day X 10 iv) or the combination of BCNU (150 mg/m2 iv) plus vincristine (VCR) (2 mg/m2 iv on Day 1 only). Treatment failures were crossed over to the alternate therapy. Primary, secondary, and cumulative response rates to DTIC were 29%, 9%, and 22%, respectively. Primary, secondary, and cumulative response rates to BCNU plus VCR were 23%, 29%, and 25%, respectively. Five of 26 patients (19%) experienced objective regression from secondary therapy after failure to respond to primary therapy. DTIC produced gastrointestinal and hematologic toxic effects; BCNU plus VCR produced gastrointestinal, hematologic, and neurologic toxic effects. VCR administered at a dose of 2 mg/m2 resulted in excessive neurologic toxic effects in 12 of 21 patients; a maximum VCR dose of 2 mg/injection was well tolerated by 15 subsequent patients without an adverse effect upon response rate. An analysis of tumor burden and organ involvement in responders and nonresponders suggests that DTIC is the first-choice treatment for patients with limited tumor burdens and nonvisceral metastases; BCNU plus VCR is the first-choice treatment for patients with extensive tumor burdens and visceral-predominant disease. However, failure to respond to primary therapy does not preclude response to secondary therapy with the alternate regimen.