The present study examines the sex-specific patterns of mortality by birth order in four stages of the life-course, using Poisson and logistic regression analysis. The main question posed is whether there is any continuing social effect of birth order when (a) biological factors at birth, (b) other social factors at birth and (c) socio-economic circumstances in adulthood are adjusted for. The analyses are based on the Uppsala Birth Cohort Study consisting of all 14,192 boys and girls who were born alive at the Uppsala Academic Hospital in Sweden during the period 1915-9. The results showed that all-cause mortality differed according to birth order in all of the four studied age intervals when birth year, mother's age, birth weight, gestational age, diseases of mother, diseases of the infant, social class and mother's marital status at the time of childbirth were adjusted for. The general tendency was for laterborn siblings, particularly girls women, to demonstrate a higher mortality risk than firstborn children. However, in the oldest age group (55-80 years) the previously significant association between birth order and male mortality became insignificant when adult socio-economic circumstances were controlled for. This indicates that the long-term influence of childhood birth order position on mortality is partly mediated by adult social class, education and income. The concluding section of the paper notes that laterborn children, and especially girls, were a disadvantaged group in early 20th century Sweden. Thus, for the subjects in the present study, the childhood social conditions linked to birth order position seem to have had consequences for these individuals' health and survival that extend over the whole life-course.