The labor supply of older men increased from the 1930s to the 1950s cohort. This paper investigates the role of three Social Security reforms in explaining these differences: a delayed normal retirement age, increased delayed retirement credits, and a change in the earnings test that was eliminated beyond the retirement age, and it evaluates the effects of several additional Social Security reforms on individuals’ behaviors. I develop and estimate a rich dynamic life-cycle model of labor supply, savings, and Social Security application that captures the key structure of the Social Security retirement benefits, disability insurance, and pension system, while taking into account uncertainties in health, survival, wages, and medical expenditures. The model is estimated using the Method of Simulated Moments, and it matches well the observed life-cycle profiles of employment, hours worked by workers, and savings for healthy and unhealthy men in the 1930s birth cohort from the Panel Study of Income Dynamics data. It shows that the three changes in Social Security rules jointly account for over 73% of the observed rises in labor force participation and hours per worker by the 1950s cohort, with the reform to the Social Security earnings test being the most important. Additional policy experiments suggest that fully eliminating the earnings test and reducing retirement benefits by 23% would further increase older-age participation by 3.4% and 5.1%, respectively.