OBJECTIVE To compare the abilities of two methods for rapid detection of pyuria and three methods of urinalysis to predict significant bacteriuria in symptomatic ambulatory women. METHODS Prospective simultaneous comparison of the results of dipstick urinalysis, standard microscopic urinalysis, and hemocytometric cell counting and Gram staining with the results of a standard urine culture. METHODS Two outpatient ambulatory care facilities serving predominantly minority and uninsured individuals. METHODS Nonpregnant women presenting with symptoms of urinary tract infection without symptoms of vaginal infection. RESULTS 105 women with symptoms were evaluated. The sensitivities of the dipstick urinalysis and the microscopic urinalysis in predicting pyuria as defined by hemocytometry were 0.76 and 0.77, respectively, and their specificities were 0.94 and 0.97, respectively. The sensitivities and specificities of the three methods of urinalysis in predicting greater than or equal to 10(4) colony-forming units (CFU)/mm3 in a urine culture were 0.88 and 0.70 for the leukocyte esterase-nitrite dipstick urinalysis, 0.98 and 0.68 for the standard microscopic urinalysis, and 1.00 and 0.49 for the Gram staining and hemocytometric cell counting. CONCLUSIONS The standard urinalysis was the most accurate single method to predict significant bacteriuria in symptomatic ambulatory women. Sequencing the dipstick urinalysis with the standard urinalysis may be a cost-effective approach to evaluating these patients in clinical practice.