Urinary tract infection (UTI), a relatively common cause of fever in infancy, usually consists of pyelonephritis and may cause permanent renal damage. This study assessed (1) the prevalence of UTI in febrile infants (temperature > or = 38.3 degrees C) with differing demographic and clinical characteristics and (2) the usefulness of urinalysis in diagnosing UTI. We diagnosed UTI in 50 (5.3%) of 945 febrile infants if we found > or = 10,000 colony-forming units of a single pathogen per milliliter in a urine specimen obtained by catheterization. Prevalences were similar in (1) infants aged < or = 2 months undergoing examination for sepsis (4.6%), (2) infants aged > 2 months in whom UTI was suspected, usually because no source of fever was apparent (5.9%), and (3) infants with no suspected UTI, most of whom had other illnesses (5.1%). Female and white infants had significantly more UTIs, respectively, than male and black infants. In all, 17% of white female infants with temperature > or = 39 degrees C had UTI, significantly more (p < 0.05) than any other grouping of infants by sex, race, and temperature. Febrile infants with no apparent source of fever were twice as likely to have UTI (7.5%) as those with a possible source of fever such as otitis media (3.5%) (p = 0.02). Only 1 (1.6%) of 62 subjects with an unequivocal source of fever, such as meningitis, had UTI. As indicators of UTI, pyuria and bacteriuria had sensitivities of 54% and 86% and specificities of 96% and 63%, respectively. In infants with fever, clinicians should consider UTI a potential source and consider a urine culture as part of the diagnostic evaluation.