The aim of this nationally-based, matched case-control study was to assess the impact of birth by caesarean section on intrapartum, and neonatal mortality among twins weighing 1500-2499 g, born in Sweden between 1973 and 1983. By using data held at the National Medical Birth Registry, Stockholm, 91 such pregnancies (study cases) where one or both twins died were identified. For each case, two controls (in all 182 pregnancies) were allotted at random from the rest of the twin pregnancies, with similar birthweight (+/- 100 g) and year of delivery (+/- 1 year). The number of twins that died was reduced from 73 during the first four years to 22 between 1977 and 1980, and to 6 during the last 3 years of the study period. Almost a quarter (23.1%) had a lethal malformation. The caesarean section rate increased during the study period, but did not differ between cases and controls (chi 2 = 1.0; P greater than 0.05). The analysis could not confirm a significant difference between cases and controls regarding the number of infants born vaginally in non-vertex presentation (chi 2 = 0.1; P greater than 0.05). The results of this study appear to indicate that birth by caesarean section was not a major factor related to the improved fetal outcome.