We combine 3 surveys that cover individuals aged 50 and over in 15 OECD countries. We then use variation in the timing of educational reforms across these countries to investigate the effects of education on health outcomes. Using statistical models that control for possible reverse causality, we find evidence that in several countries, more years of education lead to lower probabilities of self-reported poor health or functional status, and to lower prevalence of diabetes. These effects are larger than without controlling for possible reverse causality. The relationship between education and cancer is positive using either model, while the causal impacts of education on other chronic conditions are not established and require further investigation.