This study uses record linkage data for all Finnish women aged 35-64 years to examine how different measures of socio-economic status are related to mortality, and how these relationships vary according to employment status, marital status, motherhood, or age. Socio-economic mortality differentials among Finnish women exist for all cause-of-death groups analysed in this study. Although important discrepancies were observed, the patterns of mortality differentials were broadly similar by education, occupational status, family disposable income, and housing tenure. The strength of mortality differentials, however, varied according to the measure of socio-economic status used; mortality differentials by family disposable income were very small when compared with other measures of socio-economic status. Multivariate analyses by education and occupational status indicated very strong differentials in mortality from cancers (except breast cancer), and, especially, circulatory diseases according to education. Cause-specific mortality differentials by education were, however, similar in categories defined by occupational status. Relative socio-economic mortality differentials were broadly similar for most causes of death in sub-groups defined by parental status, economic activity, and marital status, with the exception of single women who experienced very large mortality differentials by education. The results indicate that no single indicator of socio-economic status is ideal. Different indicators are by no means interchangeable, and each may depict a distinct aspect of socio-economic status. Some indicators are, however, unsuitable for specific purposes. For example, researchers should hesitate to use income as a socio-economic indicator, when they are interested in causal analysis; in a middle-aged population the causal order of income, health, and mortality is unclear. Income, as well as some other indicators, is also very difficult to obtain for all persons on the basis of equal and relatively simple criteria. In order to understand how socio-economic status affects mortality it is potentially very useful to use several indicators simultaneously.