We randomized 64 patients with a history of recurrent urinary tract infections among 3 regimens of long-term (1 year) prophylactic treatment: 20 were given 100 mg. trimethoprim at night, 25 received 1,000 mg. methenamine hippurate every 12 hours and 19 were asked to cleanse the perineum (especially the periurethral area) twice daily with povidone-iodine solution. The progress of patients in terms of urinary symptoms and/or bacteriuria, changes in periurethral flora, side effects, and hematological and biochemical profiles was followed at regular intervals. All treatments were effective in reducing the incidence of symptomatic attacks when compared to the 12 months immediately before therapy and there was little to choose between the individual regimens on this account. However, trimethoprim was tolerated better than were the other 2 treatment regimens. In the group given trimethoprim most of the breakthrough infections (71.4 per cent) that occurred were caused by trimethoprim-resistant organisms (usually Escherichia coli), while in the other 2 groups the incidence of trimethoprim-resistant organisms causing infection was low (2.7 per cent). Treatment with trimethoprim reduced significantly the periurethral colonization of Escherichia coli.