High-quality data on state-level inequality and incomes, panel data on corruption convictions, and careful attention to the consequences of including or excluding fixed effects in the panel specification allow us to estimate the impact of income considerations on the decision to undertake corrupt acts. Following efficiency wage arguments, for a given institutional environment the corruptible employee's or official's decision to engage in corruption is affected by relative wages and expected tenure in the public sector, the probability of detection, the cost of fines and jail terms, and the degree of inequality, which indicate diminished prospects facing those convicted of corruption. In US states over 25 years we show that inequality and higher government relative wages significantly and robustly produce less corruption. This reverses other findings of a positive association between inequality and corruption, which we show arises from long-run joint causation by unobserved factors.