BACKGROUND: Phase 1 studies demonstrated evidence of recombinant human IL-18 (rhIL-18)-mediated immunomodulatory and clinical activity, and defined a biologically active dose range. METHODS: A phase 2 study of rhIL-18 was conducted in untreated AJCC stage IV melanoma. Patients were randomized to 1 of 3 dose groups (0.01, 0.1, and 1.0 mg/kg/d) of rhIL-18 administered as 5 daily intravenous infusions repeated every 28 days. A 2-stage design with a stopping rule was used. RESULTS: A total of 64 patients (median age, 57.5 years) with metastatic melanoma (M1a/b (30), M1c (34)) were accrued to stage I, and randomized to 3 groups (21 [0.01 mg/kg/d], 21 [0.1 mg/kg/d], 22 [1.0 mg/kg/d]). Five patients experienced 10 grade 3 drug-related adverse events (AEs): polyarthritis (1 subject: 0.01 mg/kg); deep vein thrombosis, pulmonary embolism (1:0.01 mg/kg); cognitive disorder (1:0.1 mg/kg); fatigue, dyspnea, pleural effusion, lymphopenia (1:1.0 mg/kg); fatigue, lymphopenia (1:1.0 mg/kg). One patient experienced a grade 4 AE of increased lipase (0.1 mg/kg) that led to permanent discontinuation from the study. Among 63 subjects evaluable for response, 1 (M1c; 0.01 mg/kg) achieved a partial response after 4 cycles. Four subjects (3 at 0.01 mg/kg and 1 at 1.0 mg/kg) had stable disease maintained for 6 months or longer. Due to the low apparent level of clinical efficacy using RECIST criteria, the study was terminated at the end of stage 1. The median progression free survival for the 3 groups was 7.5 (0.01), 7.4 (0.1), and 7.3 (1.0) weeks. CONCLUSIONS: rIL-18 as tested in this trial was well tolerated, but had limited activity as a single agent in patients with metastatic melanoma.