Objective. To assess morbidity and mortality in twin pregnancy deliveries, according to chorionicity and mode of delivery. Design. Population-based retrospective cohort. Setting. Fourteen obstetric departments in Denmark. Population. One thousand one hundred and seventy-five twin pregnancies with two live fetuses at 36 +0 weeks of gestation. Methods. Pregnancy outcomes assessed according to chorionicity and mode of delivery. Main outcome measures. Poor outcome defined as five min Apgar score ≤ 7, umbilical artery pH < 7.10, admission to neonatal unit for more than three days or death. Results. Dichorionic (DC) twins, delivered after 36 gestational weeks, with intended vaginal delivery (n= 689) compared with DC twins with planned cesarean section (n= 371) had an increased risk of poor outcome [odds ratio (OR) 1.47, p= 0.037] after adjustment for body mass index, parity and weight discordance. There was no increased risk for poor outcome in monochorionic (MC) twins with intended vaginal delivery (n= 63) compared with planned cesarean section (n= 52; OR 0.87, 95% confidence interval 0.26-2.96). Nulliparity increased the risk of poor outcome in DC (OR 1.5, p= 0.03) and in MC twins (OR 4.01, p= 0.02), as well as birthweight discordance >300 g (DC, OR 1.50, p= 0.02; and MC, OR 6.02, p= 0.002). For DC twins, we found a significantly higher risk of poor outcome of the second-born twin compared with the first (OR 1.64, p= 0.001). Conclusions. Dichorionic twins born after 36 weeks of gestation had a higher risk of poor outcome by intended vaginal delivery than by planned cesarean section. For MC twins, statistical differences in outcome by mode of delivery could not be seen.