The aim of this research was to study fetal and infant mortality in Sweden between 1973 and 1996 in twins vs singletons in relation to gestational duration. Analysis was of fetal and infant mortality based on the number of pregnancies at risk as the denominator rather than the number of deliveries each week. The analysis was based on information stored at the Medical Birth Registry (MBR), the National Board of Health and Welfare, Stockholm. The MBR keeps records on virtually all pregnancies (> 99%) regarding delivery and neonatal information, and for infant mortality up to 1 year of age. During the study period, 2,206,738 singleton and 52,658 twin births were registered. Risk evaluation was made as odds ratio (OR) with a 95% confidence interval. The material was stratified according to parity, maternal age, year of delivery, and delivery unit. Results showed the OR for twin births before 34 weeks gestation was 6 to 8-fold increased compared with singletons. The OR for fetal mortality was increased in all gestational weeks, and like-sexed twins had a consistently poorer prognosis compared to unlike-sexed. Between 1989–96, unlike-sexed twins had a fetal mortality approaching that of singletons. In conclusion, real progress in reduction of infant mortality in twins may be impossible until the high incidence of preterm births can be decreased. Hypothetically, about 100 twin labors would have to be induced to avoid one fetal death in like-sexed twin pregnancies.