Abstract Whether increases to statutory retirement ages will have the anticipated effect in countries of the Organization for Economic Co-operation and Development (OECD) depends on whether workers have the health capacity to postpone retirement. Methods that were proposed to measure the capacity to work at older age are ill-designed to assess inter-cohort trends, which is important for determining whether the amount of years spent retired and in good health is keeping up with increases in the statutory retirement age. We propose to measure the capacity to work as the amount of time that people would spend working if they kept working until poor health forces them to retire. We find that American workers born in 1936–1947 spent 9.4 years working between ages 55 and 69 but had the capacity to work an additional 3.3 years. We further find significant inter-cohort increases in the years spent working but insignificant ones in the years spent able to work, which might point toward a decrease in the years spent retired and in good health. Increases in the educational attainment of younger cohorts have had a positive effect on the capacity to work, but the expansion of obesity a negative one. Finally, we find similar trends among men and women as well as among Whites and non-Whites, although the capacity to work is much lower among non-Whites. Our results show the importance of considering inter-cohort changes in the capacity to work when designing policies that aim at inducing higher retirement ages.