This study investigated the impact of combining marital, parental, and occupational roles upon 18-year risk of mortality from all causes. The respondents were 3,700 participants in the Human Population Laboratory cohort ages 35-64 who completed a comprehensive health and psychosocial questionnaire in 1965 and were followed for mortality status through 1982. Employment status and type of employment were not found to predict mortality risk among women. Contrary to the multiple roles hypothesis, there was virtually no impact upon mortality of increasing numbers of children among employed women, except possibly among single working parents. The major impact of children was felt by housewives who had significantly elevated risks when a child was present in the home or when they had four or more children. Neither the number of children nor the presence of a child in the home affected mortality risk of men. Controlling for a variety of factors thought to be related to mortality in a logistic regression analysis did not change the foregoing relation.