A prospectively randomised controlled clinical trial of adjuvant therapy was undertaken, at a single-centre, population-based cancer institute, in patients with Dukes' stages B2 and C colorectal carcinoma after curative surgery. Between 1976 and 1983, 253 patients were randomised to either control (no further therapy after surgery), immunotherapy (oral bacille Calmette-GuCrin [BCG] 120 mg once a month) for 5 years or chemoimmunotherapy (oral BCG as above with methyl-cyclohexyl-chloroethyl nitrosourea [meCCNU] 130 mg/m2 on day l and 5-fluorouracil [5-Fu] 325 mg/m2/day on days 1-5 and 375 mg/m2/day on days 36-40) repeated every 10 weeks for 8 cycles. The median follow-up of patients is now 6.95 years. Of the control, immunotherapy , and chemoimmunotherapy groups 22.35%, 39.28%, and 28.57%, respectively, have relapsed. The log-rank analysis of results shows no disease-free or overall survival advantage for patients receiving adjuvant therapy compared with the control group. Patients receiving adjuvant immunotherapy for stage B2 appear to have a significantly inferior disease-free survival compared with other groups, but their overall survival is similar. There are no significant differences in disease-free or overall survival in the three groups of patients with stage C tumour. Of 82 patients dying, 78.05% died of progressive colorectal carcinoma, 13 patients developed a second malignancy; the remainder died of seemingly unrelated causes.