Gemcitabine and treosulfan are DNA-damaging agents. Preclinical studies suggest that synergism exists when melanoma cells are exposed to both drugs concurrently. We conducted a phase I trial in advanced melanoma patients to determine the optimal dose of gemcitabine to be combined with treosulfan. Cohorts of three patients received increasing doses of gemcitabine, commencing at 0.5 g m À2 , followed by a fixed dose of 5.0 g m À2 treosulfan on day one of a 21-day cycle. Patients alternately received a first cycle of single-agent gemcitabine or treosulfan before subsequent cycles of both drugs. Peripheral blood lymphocytes were collected in cycles 1 and 2 at various time points until 48 h post-treatment. The single-cell gel electrophoresis (Comet) assay was used to measure chemotherapy-induced DNA damage. A total of 27 patients were enrolled, no objective responses were observed, but two uveal melanoma patients had minor responses. Dose-limiting myelosuppression was reached at 3.0 g m À2 gemcitabine. DNA single-strand breaks were detected 4 h post-gemcitabine, repaired by 24 h. DNA interstrand crosslinks were detected 4 h post-treosulfan, fully removed by 48 h. Following combination chemotherapy, treosulfan-induced DNA crosslinks persisted, still being detectable 48 h post-treatment, supporting the hypothesis that gemcitabine potentiates treosulfan-induced cytotoxicity. The recommended regimen for further study is 2.5 g m À2 gemcitabine combined with 5.0 g m À2 treosulfan.