OBJECTIVE. The purpose of this study was to compare the usefulness of renal sonograms obtained 6 days and 6 weeks after birth in differentiating obstruction from nonobstruction in patients with antenatal pyelocaliceal dilatation shown by sonography and to establish sonographic criteria to determine the degree of postnatal pyelocaliceal dilatation that warrants further investigation. MATERIALS AND METHODS. Criteria for an infant to enter the study were fetal pyelectasis of 4 mm or greater, two postnatal sonograms with the second showing persisting pyelectasis extending at least into the infundibula, and a voiding cystourethrogram showing normal findings. One hundred thirty kidneys in 100 infants met the study criteria. The first postnatal sonogram was obtained at a mean age of 6 days (range, 1-14 days) and the second at a mean age of 6.6 weeks (range, 3-16 weeks). The degree of pyelectasis was measured in the anteroposterior direction on the transverse postnatal sonograms. The diagnosis of obstruction was made by excretory urography in 99 infants and nephrostography in one infant. Kidneys were categorized as definitely obstructed, possibly obstructed (anatomic features of obstruction on excretory urogram but functionally not obstructed), or not obstructed. Receiver-operating-characteristic (ROC) curves based on renal pelvic diameters were plotted for both sonograms; the ability to detect definite obstruction or possible obstruction was compared for the two time periods; and optimal cutoff points were determined. RESULTS. The mean diameter of the renal pelvis was not significantly different between the sonogram obtained at 6 days and the sonogram obtained at 6 weeks for the 86 nonobstructed kidneys. For the 27 kidneys that were obstructed, the mean pelvic diameter increased from 18 mm (range, 5-54 mm) on the sonogram obtained at 6 days to 22 mm (range, 11-60 mm) on the sonogram obtained at 6 weeks. The mean pelvic diameter of 17 kidneys categorized as possibly obstructed increased from 6 mm (range, 0-11 mm) to 10 mm (range, 6-20 mm) between the first and second sonograms. The ROC curves for 811 sonograms obtained at 6 weeks provided cutoff points with greater sensitivity and specificity than did the curves for the sonograms obtained at 6 days. The optimal cutoff points were 6 mm for possible obstruction (sensitivity, 100%; specificity, 57%) and 11 mm for definite obstruction (sensitivity, 100%; specificity, 96%). CONCLUSION. Renal obstruction may be underestimated or missed on a renal sonogram obtained 6 days after birth. A sonogram obtained 6 weeks after birth is more specific for detecting obstruction.