ABSTRACT.The effects of various social indicators on infant and child mortality were studied in Sweden with the use of a medical birth register to which census information was linked. Two years were studied: 1976 births linked to the 1975 census, and 1981 births linked to the 1980 census. Survival was followed to the age of 5 by linkage of the birth register with the death certificate register. The only statistically significant effect of a single socio‐economic variable was that of housing conditions on perinatal death rate and postperinatal death rate up to the age of one. The family situation (e. g., cohabitation or not) had some effect, although it was not statistically significant. On the basis of cohabitation status and other social indicators, including housing conditions, we selected two groups: one privileged and the other underprivileged. Using crude mortality rates, we found no definite difference. There was evidence that the mortality rate had decreased more between 1976 and 1981 in the privileged than in the underprivileged group, but the difference may have been coincidental. After standardization for maternal age and parity, however, a difference appeared with a ratio of 1.14 between the underprivileged and the privileged groups, which was valid for deaths up to the age of one. After that age, no difference was seen. Following standardization for birthweight, the opposite was found: a higher weight‐specific mortality rate in the privileged group than in the underprivileged group. The interpretation of these findings is discus