A double-blind, collaborative, randomized trial was initiated in 1976 to evaluate dexamethasone administered to mothers as a method of preventing neonatal respiratory distress syndrome (RDS). Five centers enrolled 696 women at risk for premature delivery. Up to 20 mg of dexamethasone phosphate (5 mg every 12 hours) or a placebo was administered intramuscularly. The overall incidence of RDS was different between control subjects (18.0%) and steroid-treating mothers (12.6%) (P less than 0.05). The effect was, however, mainly attributable to discernible differences among singleton female infants (P less than 0.001), whereas no treatment effect was observed in male infants (P = 0.96). Non-Caucasians were improved whereas Caucasians showed little benefit. Fetal and neonatal mortality and maternal postpartum infection rates were not different. Neurologic examination at 40 weeks' gestation demonstrated no significant difference in the rate of abnormal outcomes in the neonatal steroid group (P = 0.2).