The emergence of multidrug-resistant (MDR) uropathogens is making the treatment of urinary tract infections (UTIs) more challenging. We sought to evaluate the accuracy of empiric therapy for MDR UTIs and the utility of prior culture data in improving the accuracy of the therapy chosen. The electronic health records from three U.S. Department of Veterans Affairs facilities were retrospectively reviewed for the treatments used for MDR UTIs over 4 years. An MDR UTI was defined as an infection caused by a uropathogen resistant to three or more classes of drugs and identified by a clinician to require therapy. Previous data on culture results, antimicrobial use, and outcomes were captured from records from inpatient and outpatient settings. Among 126 patient episodes of MDR UTIs, the choices of empiric therapy against the index pathogen were accurate in 66 (52%) episodes. For the 95 patient episodes for which prior microbiologic data were available, when empiric therapy was concordant with the prior microbiologic data, the rate of accuracy of the treatment against the uropathogen improved from 32% to 76% (odds ratio, 6.9; 95% confidence interval, 2.7 to 17.1; P < 0.001). Genitourinary tract (GU)-directed agents (nitrofurantoin or sulfa agents) were equally as likely as broad-spectrum agents to be accurate (P ؍ 0.3). Choosing an agent concordant with previous microbiologic data significantly increased the chance of accuracy of therapy for MDR UTIs, even if the previous uropathogen was a different species. Also, GU-directed or broad-spectrum therapy choices were equally likely to be accurate. The accuracy of empiric therapy could be improved by the use of these simple rules.T he incidence of infections caused by multidrug-resistant (MDR) Gram-negative bacterial uropathogens is increasing among both hospitalized patients and patients in the community (1, 2). The emergence of these pathogens creates challenges for physicians when choosing empiric treatment, as therapeutic options are often limited. Current guidelines for the treatment of urinary tract infections (UTIs) from the Infectious Diseases Society of America (IDSA) recommend that antimicrobial resistance rates be considered and that patient risk factors be taken into account (3). However, with rates of extended-spectrum beta-lactamase (ESBL) production among uropathogens approaching 10% (4, 5), additional strategies for selecting an accurate antimicrobial agent are needed.Studies suggest that empiric treatment for patients with bacteremia caused by ESBL-producing strains accurately covers the pathogen in only half of all cases (6, 7). Inaccurate therapy has been associated with both increased morbidity and increased mortality (6-8). In addition, use of inaccurate therapy for MDR UTIs can increase both the cost of care and the length of stay for hospitalized patients (9). As urinary sources are one of the most common sources of bacteremia caused by Gram-negative bacteria and treatment for UTIs is usually initiated prior to the availability of microbiologic data,...