Better perinatal care has led to better survival of very preterm children, but may or may not have increased the number of children with cerebral and pulmonary morbidity. We therefore investigated the relationship between changes in perinatal care during one decade, and short-term outcome in very preterm infants. Perinatal risk factors and their effects on 28-day and in-hospital mortality, and on intraventricular haemorrhage and bronchopulmonary dysplasia (BPD) in survivors, were compared in two surveys of very preterm singleton infants in the Netherlands. Between 1983 and 1993, 28-day mortality decreased from 52.1% to 31.8% in infants of 25-27 weeks' gestation and from 15.2% to 11.3% in infants of 28-31 weeks' gestation. The incidence of intraventricular haemorrhage in survivors did not change (44.4% and 43.3% in infants of 25-27 weeks' gestation, and 29.0% and 24.0% in infants of 28-31 weeks' gestation). The incidence of BPD in survivors increased from 40.3% to 60.0% in infants of 25-27 weeks' gestation and remained similar in infants of 28-31 weeks' gestation (8.5% and 9.8% respectively). In multivariable analysis, higher mortality was associated with congenital malformation, low gestational age, low birthweight, no administration of steroids before birth, low Apgar scores and intraventricular haemorrhage, in 1983 as well in 1993, and with male gender in 1993. The effect of maternal age on mortality diminished significantly between 1983 and 1993. Intraventricular haemorrhage in surviving children was associated with low gestational age and artificial ventilation, both in 1983 and in 1993. The effect of artificial ventilation on the incidence of intraventricular haemorrhage diminished significantly between 1983 and 1993. BPD was associated with low gestational age and artificial ventilation, both in 1983 and in 1993, and with low birthweight and caesarean section in 1993. We conclude that the better survival of very preterm infants, especially of those of 25-27 weeks' gestation, has been accompanied by a similar incidence (and thus with an increased absolute number) of children with intraventricular haemorrhage and by an increased incidence of children with BPD.