In a prospective study of low-dose antibacterial prophylaxis of childhood urinary tract infection (UTI), co-trimoxazole and trimethoprim (TMP) have been compared for efficacy in preventing UTI, for their effect on the rectal flora and for secular selection of TMP-resistant organisms. Between 1979 and 1986, 334 children who had proven infection of an unobstructed urinary tract complied in a regimen of low-dose prophylaxis together with measures to eliminate residual urine for at least 6 months. Of these children, 167 had vesico-ureteric reflux and 27 had renal scarring. There was no difference between the two drugs in compliance, which was very good, or in the occurrence of side-effects, which were minimal. Recurrence rates of further infection were 1 per 22 child years for the 226 children receiving cotrimoxazole and 1 per 18 child years for the 108 receiving TMP. All but one of these urinary pathogens were resistant to TMP and reinfection of the urinary tract generally occurred following lapses in attention to complete bladder emptying. Neither a secular increase in recurrent infections during this period, nor a significant change in the proportions of TMP-resistant faecal coliform organisms, was observed. TMP and co-trimoxazole appeared to be equally effective prophylactic agents.