Babies with hydronephrosis detected antenatally who were born at or referred to our hospital from 1990 to 1995 were followed up with ultrasound (U/S), micturating cystourethrogram (MCU) or nuclear medicine studies after birth. One hundred and three patients were diagnosed antenatally at 17–42 weeks gestation. Twelve cases were excluded from the analysis of the results because of incomplete data. Fifty-one (56%) patients had hydronephrosis without organic obstruction, and 80% of these became normal in 3 years. Fifteen patients (17%) had a normal scan 4 days after birth. This suggests the possibility of antenatal spontaneous regression. Seven (8%) had a ureterocele and 4 (5%) had pelviureteric junction (PUJ) obstruction. Four (5%) had vesicoureteric reflux, and 4 (5%) had primary megaureter. Two (2%) had posterior urethral valves (PUV), 3 (3%) had refluxing primary megaureter, and 1 (1%) had urethral atresia. Fifteen patients (17%) underwent surgical intervention. Six had a nephrectomy, 1 a vesicostomy, 3 an Anderson-Hynes pyeloplasty, 3 had the ureterocele unroofed, 1 had a ureteric reimplant, and 1 ablation of valves. In 42 infants with 60 abnormal kidneys, the renal anteroposterior diameter of the pelvis was measured. Retrospectively, 48 kidneys diagnosed as having hydronephrosis, antenatally had a renal pelvis diameter ≥4 mm before 33 weeks gestation or ≥7 mm after 33 weeks gestation. One patient with PUJ obstruction lost kidney function, but there is no good marker to detect these patients. Early unroofing of ureteroceles may rescue kidney function. Our follow-up protocol for antenatal hydronephrosis is U/S at 4 days, 1 month and 1 year of age. An MCU is not required unless the ureter is seen on antenatal U/S. If dilatation persists past 1 month, a radionucleotide (MAG3) scan and repeat U/S are performed at 3 months. The methods for assessing obstruction and the indications for surgical intervention in these patients require reexamination.