We have looked at the outcome of pregnancy in women with type 1 (insulin-dependent) diabetes mellitus who were confined at Glasgow Royal Maternity Hospital between 1971 and 1984. Of 134 pregnancies, 123 proceeded beyond 28 weeks' gestation and yielded 125 live infants. The perinatal mortality rate was 16/1000, due solely to congenital foetal malformation. Malformations were not related either to the severity of maternal diabetes (as graded by the White classification) or the glycaemic control in the first trimester (as judged by maternal glycosylated haemoglobin concentration). Over the period of study, there has been a marked reduction in the frequency of ketoacidosis during pregnancy and in the delivery of small-for-dates babies; more mature lecithin:sphingomyelin ratios have been obtained by amniocentesis; and better Apgar scores have been demonstrated in the infants at delivery. The Caesarean section rate has fallen from 83 to 30 per cent, and babies now spend less time separated from their mothers in paediatric units. These improvements largely reflect better diabetic treatment (improved insulin regimens and glycaemic control) and closer obstetric assessment. Foetal malformation occurred overall in 11.4 per cent of pregnancies and did not become less frequent over the period of study. Further major improvement in the outcome of diabetic pregnancy will only come from a reduction in the congenital malformation rate, which implies better diabetic control at and around the time of conception.