Data from a prospective British birth cohort study showed that women who were childless, lone mothers or full-time homemakers between the ages of 26 and 53 were more likely to report poor health at age 54 than women who occupied multiple roles between these ages. To explain this finding we developed and tested a theory of role quality based on the concept of agency by drawing on Giddens' theory of structuration and Doyal and Gough's theory of human needs. According to our theory, the patriarchal structuration (drawing on Giddens' term) of work and family roles provides both limitation and opportunity for the expression of agency. Doyal and Gough's theory of human needs was then used to identify the restriction of agency as a possible influence on health. This theory of role quality was operationalised using a measure of work (paid and unpaid) quality at age 36 and a measure of work and family stress between ages 48 and 54. The relatively poor subjective health in mid-life of lone mothers was explained by work and family stress and adult social class. In contrast, the poor health in mid-life of long-term homemakers and childless women was less easily explained. Homemaker's excess risk of reporting poor health at age 54 remained strong and significant even after adjusting for role quality and socioeconomic indicators, and childless women were at an increased risk of reporting poor health despite the social advantage inherent in attaining educational qualifications and occupying professional or managerial occupations. This study highlights the need to develop measures of role quality specifically designed to capture agency aspects of social roles.