The aim of this randomized four-arm phase III study was to evaluate whether there is a difference in activity between regimens containing dacarbazine and regimens without dacarbazine in metastatic melanoma, whether there is a dose–effect relationship for dacarbazine, and whether non-dacarbazine-containing aggressive regimens are in any way superior to non-aggressive ones. A total of 219 patients with metastatic cutaneous melanoma were included in this study; 196 of them were evaluable for activity. The patients were randomized into four treatment arms: arm A (standard dose dacarbazine arm), vincristine 1.4 mg/m2 on day 1, carmustine (BCNU) 60 mg/m2 on day 1, and dacarbazine 300 mg/m2 per 24 h on days 2–5; arm B (high-dose dacarbazine arm), vincristine and BCNU as in arm A and dacarbazine 600 mg/m2 per 24 h on days 2–5; arm C (`aggressive’ regimen without dacarbazine), vindesine 3 mg/m2 on day 1, bleomycin 7 mg/m2 per 24 h on days 1–4, and cisplatin 30 mg/m2 per 24 h on days 5–8; arm D (`non-aggressive’ regimen without dacarbazine), BCNU 100 mg/m2 on day 1 and procarbazine 90 mg/m2 per 24 h on days 1–10. The four arms were well balanced with regard to patient- and disease-related characteristics. On an intend-to-treat basis, the response rate was 11 out of 49 (22%) in arm A, nine out of 47 (19%) in arm B, 16 out of 63 (25%) in arm C and nine out of 60 (15%) in arm D. There was a large overlap between the 95% confidence intervals and no significant differences in the response rates between the four arms. Median survival in the four treatment arms was 4, 5, 6 and 4 months, respectively, again with no significant differences. Median survival for responders (8, 11, 10 and 13 months, respectively) in all four arms was significantly longer than in non-responders (4, 3, 5 and 4 months, respectively). Arms A, B and C were significantly more toxic compared with arm D, which was for all practical purposes devoid of toxicities. The efficacy of all four regimens thus appeared comparable both in terms of response rate and survival. Responders in all four arms achieved a survival benefit. There does not seem to be a dose–effect relationship for dacarbazine in metastatic melanoma. Chemotherapy from arm D, might be well suited for ‘fragile’ or elderly patients due to the lack of toxicity.