Two contrasting hypotheses on the effects of combining marital, parental and work roles on mortality are analysed in this paper. The 'multiple role' hypothesis suggests that the effects are harmful, but the 'role accumulation' hypothesis argues that the benefits will outweigh the possible harmful effects. This paper uses record linkage data for all 35-64 year-old non-pensioned Finnish women to examine the two hypotheses. Women with all three roles of wife, mother and employee had low mortality. This, however, was a reflection of the main effects of these three variables. Only lone mothers with > 1 child--about 4% of the study population--were characterized as having a somewhat deviant mortality from what was to be expected on the basis of the main effects model. The high mortality in this group was mainly due to causes of death related to accidents and violence and circulatory diseases. Further analysis indicated that the excess mortality among lone mothers with two or more children and the lack of interactions for any other role constellation was similar in all age and educational groups. It is concluded that neither of the hypotheses on multiple roles are very relevant for the analysis of female mortality and that more attention should be devoted to understanding the contribution of possible selection effects leading to marriage and motherhood and the contribution of the 'healthy worker effect' in creating low mortality for the employed.