OBJECTIVES: To determine the best urinary bacterial concentration to diagnose urine infections. METHODS: We studied a quantitative culture of paired urine samples from children that were promptly tested together after serial dilution. The initial diagnosis of urinary tract infection made from the result of the first urine culture and subsequently modified according to the second sample result, and then the ratio of their colony counts was considered. A total of 203 children (aged 2.0 weeks to 17.7 years) were screened for urine infection in a hospital setting. RESULTS: The 36 children who had a urinary tract infection, defined as having the same uropathogen in both urine samples at concentrations within 25-fold of each other, had a mean colony count of 1.7 × 107 colony-forming units/mL. Among the 167 children who did not have a urinary tract infection, 12 (7.2%) would have had a false-positive diagnosis made on the first sample, which was revealed because the second sample result was different (n = 7) or had a ≥25-fold different colony count (n = 5). Raising the threshold from 105 to 106 colony-forming units/mL reduces the false-positive rate 4.8%. If 2 samples are cultured, the false-positive rates fall to 3.6% and 0.6%, respectively. All 9 children (5.4% of those without a urinary tract infection) who had a mixed culture with ≥105 colony-forming units/mL of a uropathogen (heavy mixed growth) in the first sample had a urine infection excluded by the second sample result. CONCLUSION: The minimum urinary bacterial concentration that is used to diagnose a urine infection should be increased from ≥105 to ≥106 colony-forming units/mL, because that would reduce the false-positive rate from 7.2% to 4.8% if 1 sample was cultured and from 3.6% to 0.6% if 2 samples were cultured. Urine samples with heavy mixed growths should be considered contaminated.