Thirty-three patients with metastatic melanoma were treated in a phase II study with an intravenous continuous infusion (IVCI) of interleukin-2 (IL2) given with lymphokine-activated killer (LAK) cells. The dose of IL2 was the optimal priming dose for LAK-cell induction, followed by the maximally tolerated LAK-cell dose that could be given by an IVCI schedule as determined by a previous phase I trial. The CI schedule was chosen for evaluation because of a postulated reduction in toxicity with the possibility of administering a more prolonged IL2 infusion and because greater rebound lymphocytosis and LAK-cell generation had been reported using this dose and schedule. The 33 patients were similar in age, performance status, and sites of disease to those treated in previous IL2 trials. All patients were assessable for response and toxicity. One patient (3%) achieved a partial response of 10 months duration. There were no other clinically significant responses. Significant toxicity included hypotension requiring pressors (45%), dyspnea (36%), renal insufficiency (24%), hepatic dysfunction (66%), and cardiac arrhythmias (18%). These toxicities reversed with cessation of the infusion. There were four deaths during the first 30 days of treatment, three from infection (one related to central line, one related to LAK cells, one related to tumor), and one from tumor-related hemorrhage. Toxicity was unexpectedly high and at least comparable to that seen in previous studies using a high-dose IV bolus schedule of IL2. When comparing the IVCI schedule with high-dose bolus IL2 to LAK cells in nonrandomized but sequential studies in patients with advanced melanoma, it appears that CI IL2 is less efficacious.