Excluding specimen contamination, bacteriuria indicates either urinary colonization (replication of bacteria in urine without evidence of tissue invasion) or urinary tract infection (bacteriuria associated with clinical, histologic, or immunologic evidence of host injury). Practically speaking, measuring pyuria is the most readily available means of establishing the presence of host injury, thus differentiating colonization from infection. Careful studies have established the nonpathologic limits for pyuria (less than 10 leukocytes/mm3 in uncentrifuged urine) and have compared methods for its measurement. The most commonly used method, determination of cells per high power field in centrifuged urine, is not reproducible and does not correlate with either the actual leukocyte excretion rate or the cells per cubic millimeter as enumerated in a counting chamber. When clinical studies using the latter method of determining pyuria were reviewed, the following conclusions emerged: (1) 10 leukocytes/mm3 or greater occur in less than 1 percent of asymptomatic, nonbacteriuric patients but in greater than 96 percent of symptomatic men and women with significant bacteriuria; (2) most symptomatic women with pyuria but without significant bacteriuria have urinary infection either with bacterial uropathogens present in colony counts less than 10(5)/ml or with Chlamydia trachomatis; (3) women with asymptomatic bacteriuria probably should be divided into two subgroups: those with true asymptomatic infection (associated with pyuria) and those with transient, self-limited bladder colonization; and (4) most patients with catheter-associated bacteriuria also have pyuria and hence infection. In view of its potential value to both clinicians and microbiologists, pyuria should be accurately determined and expressed as cells per cubic millimeter in uncentrifuged urine.