We analyzed retrospectively (1980-1990) the causes of death and by using a logistic regression model the perinatal and neonatal risk factors influencing the mortality in preterm infants < 34 weeks of gestation (n = 1132). When comparing the interval from 1980-1986 to 1987-1990 we observed a decreasing mortality in infants < or = 1000 g from 57% to 19% as well as in the preterm infants > 1000 g from 8.3% to 3.0% (p < 0.001). The causes of death changed considerably. During 1980-1986 fifty-two (8.2%) out of the 632 preterm infants and during 1987-1990 only seven (1.3%) out of the 600 preterm infants died in the course of a severe respiratory distress syndrome or intracranial hemorrhages. From 1980 to 1986 21% (n = 10) and from 1987 to 1990 77% (n = 10) of the neonatal deaths in preterm infants > 1000 g were attributed to lethal malformations. In those infants without lethal malformations (n = 1109) we performed a logistic regression analysis. 87 (7.8%) of these neonates died. The risk of dying was significantly higher in infants born before 1987, in male newborns and in infants suffered from a severe respiratory distress syndrome III-IV or septicemia (p < 0.0001). An increasing gestational age of one week resulted in a lowered risk of mortality (odds ratio 0.59, p < 0.0001). Adjusted for these basic variables the mortality risk was also significantly higher for birth weights < or = 1000 g, low Apgar scores, peripartal acidosis, hypothermia and intracranial hemorrhages. An intrauterine growth retardation < 10. percentile resulted in a lower mortality risk.