Health selection out of the labour force has received considerable attention by analysts attempting to disentangle the "true" biological dimensions of ill-health from its social meaning. Rejecting this dualistic separation, we argue that the effect of health on labour force participation is an inherently social process reflecting differential access to material and symbolic rewards that are structured by social position. Using longitudinal data from the US-based Panel Study of Income Dynamics, we examine the extent to which structural arrangements, including those designated by gender, race, education and age, differentially affect the risk of a labour market exit when health is compromised. Individuals employed at entry into the study (from 1984-1990) were followed for the duration of the study or until they left the labour force. Analyses were stratified by gender and age (25-39 and 40-61 years at baseline). We found suggestive evidence that the hazard of labour market exit in the context of perceived ill-health depended on gender, race and education, but in ways that were not constant across each of these social positions. For example, men may be more vulnerable to the labour market effects of poor health, but only in the younger group, black men were less likely to leave the labour force than white men, and education mattered, but only among younger women and older men. While these patterns may reflect differential access to disability pensions or other work-related benefits, we suggest that a more detailed analysis of trajectories of health and employment. as well as the meaning of health states would be useful in further elucidating the social dimensions of health selection.