In this paper, we examine associations between employment history and marital status and unpaid care provision among those aged 40–59 in England and Wales. We used data from a large nationally representative longitudinal study, the Office for National Statistics Longitudinal Study. Initially based on a sample drawn from the 1971 Census, in 2001 this study included data on 110 464 people aged 40–59 of whom 5% provided 20 or more hours per week of unpaid care. We analysed associations between caregiving of this intensity and current employment, employment history, employment characteristics, marital status, and employment after childbearing. Among men, caregiving was associated with a history of lower levels of employment. The small group of men with a history of least employment were 70% more likely to provide care than those with a history of most employment. Among women, caregiving was associated with a history of non-employment, but there were no differences between those with fully engaged and partially engaged labour market histories. Analyses of a subset of data on women who had a child between 1981 and 1991 showed that those who had returned to full-time paid work by 1991 were over 50% less likely to later become caregivers. Some associations between employment characteristics and propensity to provide 20 or more hours per week of care were also identified. Those in public sector jobs and those previously in employment with a caregiving dimension were 20–30% more likely than other working women to provide unpaid care. These results suggest a continuing gender dimension in care provision which interacts with marital status and employment in gender-specific ways. It also suggests that implementation of strategies to enable those in midlife to combine caregiving and work responsibilities, should they wish to do so, should be an urgent priority.