We provide the first empirical analysis of gubernatorial pay. Using US data for 1950-90 we document, contrary to widespread assumptions, substantial variation in the wages of politicians, both across states and over time. Gubernatorial wages respond to changes in state income per capita and taxes, after controlling for state and time fixed effects. The economic effects seem large: governors receive a 1 percent pay cut for each ten percent increase in per capita tax payments and a 4.5 percent increase in pay for each ten percent increase in income per capita in their states. There is strong evidence that the tax elasticity reflects a form of reward-for-performanc.' The evidence on the income elasticity of pay is less conclusive, but is suggestive of rent extraction' motives. Lastly, we find that democratic institutions seem to play an important role in shaping pay. For example, voter-initiatives and the presence of significant political opposition lead to large reductions in the income elasticity of pay, and to large increases (at least double) in the tax elasticities of pay, relative to the elasticities that are observed when these democratic institutions are weaker.