Between April 1988 and December 1990, 37 patients with progressive histologically proven metastatic melanoma were treated with interleukin-2 according to three multicentric successive protocols. Eighteen males and 19 females entered the trial. Mean age was 44 years (range 20-66); none of the patients had severe visceral disease or brain metastasis. Superficial and visceral metastatic sites were equally distributed. Interleukin-2 was administered as a 3 to 5 day continuous intravenous infusion, at a dose varying from 16 to 24 million international units/m2/day, as previously described by West. The second course was given after a 9 to 16 day free interval and one to seven courses were administered (mean three courses). A total of 132 courses has been given to the 37 patients. All are evaluable for toxicity and efficacy. Toxicity was tolerable and not different from that presented in recent reports. Only four patients had to definitively stop therapy for toxicity, one of them for cardiotoxicity; a dose modification or a transient suspension of therapy occurred in 18% of treatment cycles. One hypothyroidism with anti-thyroglobulin and anti-microsome antibodies was observed. We observed eight major responses (21.6%), usually of short duration (2-6 months). Most responses occurred in superficial lesions. One patient remains in complete remission, as therapy is stopped for 40 months. Immunological parameters, although demonstrating induced immunostimulation, did not correlate with clinical outcome. With an overall response rate of 21.6%, we confirm the activity of interleukin-2 in melanoma, as previously reported by others.