Fifty elderly (mean age, 83.4 +/- 8.8 years) institutionalized women with asymptomatic bacteriuria were randomly assigned either to receive therapy for treatment of all episodes of bacteriuria identified on monthly culture or to receive no therapy unless symptoms developed. Subjects were followed for one year. The therapy group had a mean monthly prevalence of bacteriuria 31 +/- 15 percent lower than those in the no-therapy group, but periods free of bacteriuria lasting six months or longer were documented for only five (24 percent) subjects. For residents receiving no therapy, 71 percent showed persistent infection with the same organism(s). Antimicrobial therapy was associated with an increased incidence of reinfection (1.67 versus 0.87 per patient-year) and adverse antimicrobial drug effects (0.51 versus 0.046 per patient-year) as well as isolation of increasingly resistant organisms in recurrent infection when compared with no therapy. No differences in genitourinary morbidity or mortality were observed between the groups. Thus, despite a lowered prevalence of bacteriuria, no short-term benefits were identified and some harmful effects were observed with treatment of asymptomatic bacteriuria. These data support current recommendations of no therapy for asymptomatic bacteriuria in this population.