Cisplatin and dacarbazine are used widely in the treatment of metastatic melanoma. To evaluate high-dose cisplatin and dacarbazine, 32 patients with metastatic melanoma were treated with cisplatin 50 mg/m2 and dacarbazine 350 mg/m2 daily for three days repeated at 28-day intervals. Their median age was 43.5 years (range, 25 to 73 years), and their median Karnofsky performance status was 80% (range, 70% to 100%). Measurable and evaluable disease sites (number of patients) included lymph nodes (22), lung (17), soft tissue (16), liver (13), bone (seven), spleen (four), adrenal gland (three), skin (three), and other sites (five). Patients received a median of two cycles of therapy (range, one to eight cycles). Thirty patients were evaluable for response. No complete responses were observed. Five patients had a partial response (17%; 95% confidence interval, 3% to 30%) for 16+, 12+, 7, 6.5, and 3 months. Responding sites of disease included lymph nodes (five of 22), lung (three of 17), and soft tissue (two of 16). Hematologic toxicity (Grade greater than or equal to 3) included neutropenia (16 of 32 patients, 30 of 90 cycles), thrombocytopenia (eight of 32 patients, 12 of 90 cycles), and anemia (five patients). Nine episodes of neutropenia and fever were seen in four patients; two had bacteremia. Nonhematologic toxicity (Grade greater than or equal to 3) included hypotension (two patients), nausea and vomiting (four), neuropathy (two), ototoxicity (four), and hypomagnesemia (nine). The low objective response rate and severe toxicity of this regimen preclude its standard use in patients with metastatic melanoma. A review of cisplatin-based therapy in metastatic melanoma suggests that there is no dose-response relationship. The use of high-dose cisplatin (greater than 100 mg/m2) in the treatment of metastatic melanoma is not recommended.