We report our experience with 92 antenatal amnioinfusion procedures. In order to facilitate ultrasound visualization, a diagnostic infusion was attempted at a median of 22 weeks (range 16-36) in 61 pregnancies with oligohydramnios in the absence of ruptured membranes on clinical examination. The procedure was successful in 58 (95%). Infusion (mean volume 181 mL, range 40-64) significantly increased (P less than .001) the deepest pool of amniotic fluid to a mean of 3.2 cm. Suspected fetal anomalies were then confirmed in 27 of 30 cases, whereas kidneys were clearly demonstrated in three fetuses suspected of renal agenesis. In addition, previously unsuspected anomalies were identified in five. Vaginal leakage indicating ruptured membranes occurred in 16 women. Leakage occurred in zero of 24 patients with, compared to 16 of 35 without, fetal urinary disorders (chi 2 = 15.1, P less than .001), which does not support the recent suggestion that amnioinfusion causes rather than unmasks rupture of the membranes. Membranous detachment was observed by ultrasound in 13 patients, 11 of whom leaked vaginally. Information obtained at amnioinfusion led to a change of etiologic diagnosis in eight (13% of subjects). Forty serial infusions were performed in nine women as a pilot study to prevent oligohydramnios sequelae. There were no skeletal deformities; three neonates survived, and five of the six perinatal deaths had normal lung-body weight ratios. Overall, only two of 89 infusions (2.2%) were complicated by clinical amnionitis. Our findings support a role for amnioinfusion in oligohydramnios.