Reduced glutathione (GSH) production by tumour cells has been proposed as a mechanism for resistance to alkylating agents. High levels of paracetamol can deplete intracellular GSH. We conducted a phase I trial of high dose paracetamol and carmustine (BCNU) in patients with advanced malignant melanoma to determine the optimal biological dose and the maximum tolerated dose (MTD) with the goal of increasing sensitivity to BCNU by GSH depletion. Groups of three to five patients received escalating doses of paracetamol (10, 15 or 20 g/m(2)) every 3 weeks. Every other cycle, BCNU (10 mg/m(2)) was given 6.5 h after administration of paracetamol and 45 min before a 20 h infusion of N-acetylcysteine. Once the MTD for paracetamol had been determined, the dose of BCNU was sequentially escalated in subsequent cohorts to 150 mg/m(2). GSH levels were measured in peripheral blood mononuclear cells (PBMCs) and, when available, in tumour biopsies. The MTD of paracetamol was 15 g/m(2). The dose of BCNU was safely escalated to 150 mg/m(2). The most common toxicity was grade II nausea/vomiting. At 15 g/m(2), peak paracetamol levels (median 253 microg/ml) were reached between 1 and 4 h. No changes in GSH levels in PBMCs were seen. There were two partial responses, including a dramatic decrease in hepatic metastases. Treatment of melanoma patients with paracetamol (15 g/m(2)) every 3 weeks and BCNU (150 mg/m(2)) every 6 weeks is safe. The observation of two partial responses has led to a phase II study to evaluate treatment with high dose paracetamol alone or in combination with BCNU.