Currently hospitalization for children with urinary tract infections (UTIs) is reserved for severe or complicated cases. Changes may have taken place in the characteristics and causative uropathogens of hospital-treated community-acquired UTI. To study children hospitalized in a tertiary center with community-acquired UTI, compare Escherichia coli and non-E. coli UTI, define predictors for non-E. coli UTI and elucidate the appropriate therapeutic approach. A prospective clinical and laboratory study from 2001 through 2002 in a tertiary pediatric medical center. Patients were divided by results of the urine culture into E. coli and non-E. coli UTI groups, which were compared. Of 175 episodes of culture-proved UTI, 70 (40%) were caused by non-E. coli pathogens. Non-E. coli UTI was more commonly found in children who were male (P = 0.005), who had underlying renal abnormalities (P = 0.0085) and who had received antibiotic therapy in the prior month (P = 0.0009). Non-E. coli uropathogens were often resistant to antibiotics usually recommended for initial therapy for UTI, including cephalosporins and aminoglycosides; 19% were initially treated with inappropriate empiric intravenous antibiotics (compared with 2% for E. coli UTI, P = 0.0001), with a longer hospitalization. Current treatment routines are often inappropriate for hospitalized children with non-E. coli UTI, which is relatively common in this population. The defined risk factors associated with non-E. coli UTIs and its antimicrobial resistance patterns should be considered to improve empiric antibiotic therapy for these infections.