Between november, 1974 and june, 1977, 64 adult patients with acute myeloid leukaemia in remission were divided at random into two treatment groups: chemotherapy alone (n = 32) or chemotherapy plus B.C.G. (n = 32). Fresh B.C.G. from the Pasteur Institute, Paris, was administered once a week, two or three times per month, in the intervals between chemotherapy cycles, using the Heaf-gun method. The duration of the first remission was significantly increased in the group receiving B.C.G. (median duration 685 days versus 390 days: p = 0.03 on log rank test) as was the overall survival (median duration of survival 1.471 days versus 554 days: p = 0.01). Second remissions were more frequent in the B.C.G. group (79% versus 42%; p = 0.014). In addition, prolonged survival was more frequent in the B.C.G. group, where 12/32 patients remained in first remission between 3 and 6 years, as against 5/32 in the chemotherapy group.