Seventy-nine patients with Stage III widely metastatic melanoma were prospectively randomised to a 'no treatment' control group who received on tumour progression DTIC (250 mg/m2 i.v. daily X 5) and Actinomycin D 1.5 mg/m2 on Day 1. A total of six courses at 3-week intervals was given. Chemotherapy was only given on progression of disease. The other group received initially Corynebacterium parvum (2 mg/m2) every 3 weeks for a maximum of eight courses and then the same chemotherapy on evidence of progressive disease. Minimum follow up time is 3 yr. The chemotherapy response rate (control 37%, C. parvum 24%) was not statistically different nor was the effect of chemotherapy on the site of individual metastases. Radiotherapy responses for irradiated soft tissue disease again were not significantly different, between the two patient groups. No significant differences in survival (control group median, 4 months, range 1-46; C. parvum median 3 months range 1-35) were observed and only one patient is alive at 35 months. The pattern of relapse was also similar in both groups. Reduction in haematological toxicity consequent on chemotherapy was not observed in the C. parvum-treated patients. No additional benefit was observed when C. parvum was followed by DTIC and Actinomycin D chemotherapy compared with the results from the chemotherapy given alone, although C. parvum on this schedule had minimal toxicity.