A retrospective study comparing the findings of ultrasonography (5 MHz transducer) versus intravenous pyelography and voiding cystourethrography was conducted in 92 patients during the course of an initial urinary tract infection (greater than or equal to 10(5) bacteriae/ml, WBC greater than 25/mm3) gathered over a period of 5 years. The average interval of time between ultrasonography and intravenous pyelography was 8 days (range 1-70). False positive results for ultrasonography were noted in 11% (5 of 45 cases), while false negatives were observed in 36% (17 of 47). Of the 47 cases of obstructive uropathy diagnosed by conventional means, ultrasonography missed the diagnosis in 1 case of ureteral duplication as well as in 16 children with vesicoureteral reflux. In comparison, intravenous pyelography missed 8 of these 21 cases of vesicoureteral reflux. Overall sensitivity of ultrasonography was mediocre at 64% but excellent at 96% after excluding cases with vesicoureteral reflux. Specificity was poor for the group of patients with vesicoureteral reflux (33%) and did not improve when they were excluded (40%). Sensitivity and specificity were comparable whatever the age group. In conclusion, we would recommend that when both ultrasonography and voiding cystourethrography are normal in children with a urinary tract infection, intravenous pyelography could safely be deferred.