Fifty-two patients with nonlymphocytic leukaemia were studied during remission induction treatment in a randomized trial to ascertain the effect of prophylactic oral trimethoprim-sulfamethoxazole on infection and fever rate. A decrease in the total number of acquired infections was found (16 infections in the group given trimethoprim-sulfamethoxazole versus 31 in the control group, p less than 0.01). The number of patients without any infection in the trimethoprim-sulfamethoxazole group was 13 compared to only three in the control group (p less than 0.01). Patients in the trimethoprim-sulfamethoxazole group needed parenteral antibiotics during 33% of the days they were granulocytopenic compared to 61% of these days for patients in the control group. However, six of nine bacteriologically documented infections in the trimethoprim-sulfamethoxazole group were caused by resistant microorganisms compared to two out of 20 in the control group.