To determine whether prenatal corticosteroid therapy would reduce the incidence of neonatal necrotizing enterocolitis (NEC), we assigned a total of 466 women admitted in premature labor either to receive placebo (group A, n = 256), if delivery was expected to occur within 24 hours of admission, or to receive betamethasone (group B, n = 210) if delivery was expected to take place more than 24 hours after admission. All women were free of severe medical complications or drug therapy; cases of intrauterine growth retardation or premature rupture of the membranes were excluded. Their newborn infants, excluding malformed, congenitally infected, and growth-retarded infants, were enrolled in the study unless they had died before the age of 10 postnatal days. Babies born to group A mothers (n = 248) were further assigned to a treatment group (group A1, n = 130) receiving dexamethasone, 2 mg/kg/day by intravenous injection during the first 7 days of life, or to a control group (group A2, n = 118) receiving 10% dextrose solution placebo. Group B infants (prenatal betamethasone, n = 205) received neither treatment nor placebo. The incidence of NEC in group A1 was 6.9% (9/130), and in group A2 it was 14.4% (17/118) (p less than 0.05). In group B the incidence was 3.4% (7/205); this was much lower than in group A2 (p less than 0.01) and lower than in group A combined (10.4%) (p less than 0.01). There was no death from NEC and no surgical intervention among group B patients. The mortality rate for group A1 (11%) was lower than for group A2 (56%) (p less than 0.02). There were fewer indications for surgical intervention for NEC in group A1 than in group A2. Histologic studies confirmed bowel ischemia in all specimens analyzed. These data support the hypothesis that the incidence of NEC is significantly reduced after prenatal steroid treatment. Although postnatal therapy with steroids does not decrease the incidence as effectively as prenatal therapy, it improves clinical outcome of NEC.