Forty-nine ambulatory children between 2-1/2 and 12 years of age with acute, clinically uncomplicated urinary tract infections caused by susceptible organisms were randomized to receive a single dose of amoxicillin based on weight or a 10-day course of amoxicillin therapy (conventional therapy). Patients receiving single doses of amoxicillin had a cure rate of 63%, which compares unfavorably with the cure rate of 92% in patients given conventional therapy. A failure of single-dose therapy predicted underlying radiologic abnormalities with a sensitivity of 60% and a specificity of 58%, making it a poor screening test for detecting those patients at risk for renal parenchymal damage. The antibody-coated bacteria assay had no predictive value in separating upper and lower tract disease, although it may predict underlying radiologic abnormalities. The data indicate that the response to single-dose amoxicillin therapy fails to separate upper from lower tract disease reliably and has a limited role in predicting response to conventional antimicrobial therapy.