In this paper, we specify a dynamic programming model that addresses the interplay among health, financial resources, and the labor market behavior of men in the later part of their working lives. Unlike previous work which has typically used self reported health of disability status as a proxy for health status, we model health as a latent variable, using self reported disability status as an indicator of this latent construct. Our model is explicitly designed to account for the possibility that the reporting of disability may be endogenous to the labor market behavior we are studying. The model is estimated using data from the Health and Retirement Study. We compare results based on our model to results based on models that treat health in the typical way, and find large differences in the estimated effect of health on behavior. While estimates based on our model suggest that health has a large impact on behavior, the estimates suggest a substantially smaller role for health than we find when using standard techniques. We use our model to simulate the impact on behavior of raising the normal retirement age, eliminating early retirement altogether and eliminating the Social Security Disability Insurance program.