The British tradition of analysing differences in health has been dominated by class, with women belatedly entering this debate. The American tradition has been dominated by role analysis, with women's health considered primarily in terms of their marital, parental and employment roles, with recent research coming to contradictory conclusions. Research in both traditions has reached an impasse. This paper uses a sample of over 25,000 men and women from the 1985 and 1986 British General Household Survey to show how both traditions need to be reformulated and integrated. The ways in which family roles are associated with women's health status is determined by material circumstances, but the material circumstances cannot be captured by occupational class alone. Participation in the labour market and consumption divisions, in the form of housing tenure, are crucial additional indicators of structural disadvantage. Standardised limiting long-standing illness ratios and multivariate logit analysis confirm that occupational class and paid employment are the most important attributes associated with health status for women and men. Family roles are important for women; women without children and previously married women have particularly poor health status especially those not in paid employment and living in local authority housing.