Fifty-three neonates born to 51 mothers treated with lorazepam were followed up for five days after delivery. Lorazepam had been given by mouth to 35 mothers and intravenously to 16. In general, maternal plasma concentrations of lorazepam were higher than the corresponding cord plasma concentrations. Cord plasma concentrations exceeding 45 micrograms/l were associated with three-quarters of the infants requiring ventilation at birth. Neonates conjugate lorazepam slowly to the pharmacologically inactive glucuronide, which is then excreted in the urine, where it is detectable for over seven days. Though lorazepam was detectable in breast milk, the maximum amounts that an infant could absorb would be pharmacologically insignificant. Full-term neonates whose mothers had received oral lorazepam had no complications apart from slight delay in establishing feeding, which in seven out of 29 cases was associated with relatively large doses of lorazepam. Intravenous lorazepam for severe hypertension was associated with significantly low Apgar scores, need for ventilation, hypothermia, and poor suckling. Preterm babies whose mothers had been given lorazepam by either route had a high incidence of low Apgar scores, need for ventilation, hypothermia, and poor suckling. These babies had lower Apgar scores than those whose mothers had received diazepam, but the diazepam group were heavier and more mature. Lorazepam was an effective sedative and anxiolytic by either route, and there were no eclamptic fits among the lorazepam-treated patients. The effects of lorazepam on neonates indicate that its intravenous use at any stage in pregnancy and oral use before 37 weeks should be restricted to hospitals with facilities for neonatal intensive care.