Estimates of effect of health on employment differ from study to study due to differences in methods, data, institutional background and health measure. We assess the importance of these differences, using a unified framework to interpret and contrast estimate for the US and England. We find that subjective and objective health measures, and subjective measures instrumented by objective measures produce similar estimates but only if a sufficiently large number of objective measures is used. Otherwise, objective measures produce downward biased estimates. Failure to account for initial conditions produces upward biased estimates. We find that a single subjective health index yields similar estimates to multiple measures. Overall, declines in health explain up to 15% of the decline in employment between ages 50 and 70. The effects drop with education and are larger in the US than in England. Cognition has little added explanatory power once we control for health.