Functional impairment trajectories in late adulthood over a decade are characterized using the Longitudinal Retirement History Study (LRHS) data set. Non-linear patterns of age-related increase in impairment are documented with longitudinal data. Subsets of panelists from the initial large, nationally representative probability sample of employed males and unmarried employed females (N = 11,000) also exhibit differential non-linear trajectories of impairment by sex, income, and educational attainment. Concurrent analysis of the relationship over a decade among sex, SES, and functional impairment suggests that in health research, both sex and SES are broad proxy variables whose usefulness for understanding health outcomes and for policy analysis would be enhanced by the specification of components.