One hundred and fifteen patients with histologically proven primary cutaneous melanoma were entered into a randomized prospective study following surgical removal of clinically obvious regional metastases. The study started in 1977 and was completed in June 1982. Fifty-eight patients (10 with regional skin metastases and 48 with regional node metastases) were in the control group and 57 (15 regional skin, and 42 regional node metastases) received adjuvant C. parvum. Other clinical features were well balanced for each group. C. parvum, 2 mg/m2 IV was repeated eight times at three weekly intervals. The schedule was based on previous work which indicated significant increase in killer cell activity. There was no serious toxicity and 22 patients exhibited tachyphylaxis. Seventeen patients in the control group (11 tumour free) are alive at 24-80 months, and 20 patients in the C. parvum group (12 tumour free) are alive 24-76 months from randomization. No statistically significant difference (P greater than 0.05) was observed between the two patient groups for overall survival (median 26 months) or relapse free survival (median 8 months) from the date of randomization. Nor was the overall survival from the date of primary surgery (median 52 months, range 4-237) or the relapse free survival, median 28 months, range 2-225, significantly different between the two groups of patients. The relapse pattern in the two groups, and survival after relapse were no different though the response rate to actinomycin D and DTIC chemotherapy was significantly higher in the control (48 per cent) than in the C. parvum (23 per cent) group. Toxicity to chemotherapy was not significantly affected by previous C. parvum treatment. Although there were trends for better survival, in the C. parvum patient group, statistically significant activity has not been noted in this study.