Seventy-two patients with a history of at least three attacks of urinary infection in the previous 12 months were assigned randomly to long-term prophylaxis with 100 mg at night of either Macrodantin (34 patients) or trimethoprim (38 patients). The mean interval between symptomatic attacks while on either treatment was increased three-fold compared with the pretreatment period. Macrodantin was significantly more effective (P less than 0.05) at preventing bacteriuria. Prophylaxis was equally effective in patients with and without a radiological abnormality. Side effects were significantly more common (P less than 0.05) in the group taking Macrodantin. In patients taking trimethoprim acquisition of resistance by faecal coliforms occurred at a rate of about 5%/month, and breakthrough infections were almost exclusively caused by trimethoprim-resistant coliforms. No acquisition of resistance occurred in patients taking Macrodantin, and the few breakthrough infections noted were due to sensitive bacteria.