International comparison of perinatal mortality rates is subject to various problems in methods of reporting. The main difficulties are differences in the definition of livebirth and the variation in policies on the elective delivery of very preterm fetuses at risk of intrauterine death. This study, based on geographically defined populations within the European Community, set out to compare survival rates in very low birthweight infants (below 1500 g). The first phase of the study in seven populations showed pronounced differences in survival and in the proportion of liveborn infants who weighed less than 1000 g. A further phase studied the outcome of all pregnancies from 22 to 28 weeks' gestation in four populations. There was no significant difference between the four populations in crude survival rates per 1000 livebirths among infants born at 22 to 28 weeks' gestation, but more babies were born alive at the earlier gestations in the United Kingdom. Survival rates were higher in Scotland and England when gestational age at birth was controlled for. Future comparisons of the outcome of perinatal care for extremely immature infants would be more meaningful if data on the outcome of all pregnancies ending after 22 weeks' gestation were routinely collected through the registration system.