Purpose. It was to quantify the intraobserver and interobserver variability of the sonographic measurements of renal pelvis and classify hydronephrosis severity. Methods. Two ultrasonographers evaluated 17 fetuses from 23 to 39 weeks of gestation. Renal pelvis APD were taken in 50 renal units. For intraobserver error, one of them performed three sequential measurements. The mean and standard deviation from the absolute and percentage differences between measurements were calculated. Bland-Altman plots were used to visually assess the relationship between the precision of repeated measurements. Hydronephrosis was classified as mild (5.0 to 9.9 mm), moderate (10.0 to 14.9 mm), or severe (≥15.0 mm). Interrater agreement were obtained using the Kappa index. Results. Absolute intraobserver variation in APD measurements was 5.2 ± 3.5%. Interobserver variation of ultrasonographers was 9.3 ± 9.7%. Neither intraobserver or interobserver error increased with increasing APD size. The overall percentage of agreement with the antenatal hydronephrosis diagnosis was 64%. Cohen's Kappa to hydronephrosis severity was 0.51 (95% CI, 0.33 to 0.69). Conclusion. Inter and intraobserver APD measurement errors were low in these group, but the agreement to hydronephrosis diagnosis and classification was fair. We suggest that standard and serial APD measurement can better define and evaluate fetal hydronephrosis.