DNA ploidy and S-phase fraction were measured by flow cytometry in the tumour tissue of 87 patients with disseminated malignant melanoma, who had been classified either as responders or with progressive disease in a study of the effects of 2 chemotherapeutic regimens. The patients had been randomized to receive treatment with dacarbazine (DTIC) and vindesine (Eldesine) with or without addition of cisplatin (Platinol). Tumour tissue was obtained from both the primary tumours and the last histologically verified metastases, but in some cases only the primary tumours or the last metastases could be evaluated. There was a significantly higher mean S-phase value in melanoma metastases from patients with complete or partial responses compared with patients with progressive disease. Neither the S-phase fraction of the primary tumour, nor the DNA ploidy of the primary tumour or of the last histologically verified metastases taken before inclusion into the study were associated with therapeutic response. In the multivariate analysis, both the anatomical location of the metastases and the S-phase fraction measured on the last metastases remained significant prognostic factors of response. In the univariate survival analysis, there was an association between high S-phase fractions of the metastases and longer survival. In the multivariate survival analysis, the S-phase fraction, the number of involved metastatic sites and the treatment response were independent predictive factors. We conclude that, in disseminated melanoma treated with chemotherapy, a high S-phase fraction measured in the last histologically verified metastases is associated with a higher response rate and a longer survival. Our results clearly support the role of S-phase measurement as a potential tool for selecting patients for treatment.