We employ data from the Panel Study of Income Dynamics to investigate income to health causality. To account for unobserved heterogeneity, we focus on the relationship between earnings growth and changes in self-reported health status. Causal claims are predicated upon appropriate moment restrictions and specification tests of their validity. We find evidence of Granger-type causality running from income to health for married men but not for women or single men. These effects are more pronounced for younger men and the bottom quartile of the earnings distribution. The former may be the consequence of permanent earnings shocks, whereas the latter may be the consequence of job loss.