This paper contributed to better understanding a counterintuitive, yet empirically solid, finding of recent research on Latin America: that increases in corruption have been accompanied by reductions in income inequality. By using Bayesian estimates of corruption perception suitable for panel estimations, and by identifying previously ignored meaningful differences in the typologies of corruption, we show that this effect has been driven by the consistent use of perception surveys as a proxy to measure corruption. Once detailed, experiencedbased measures of corruption are used, the marginal impact of corruption on inequality is negative when it provides of informal labor market opportunities to the general public, but positive when firms gain access to privilege treatments. By correcting measurement errors from previous fixed-effect estimations, our results show that general public-level corruption tends to increase the share of income for the bottom 20%, while firm-level corruption tends to increase the share of the top 10%. Our results indicate a pressing need to develop a more precise picture of how corruption and inequality interact, and to define how corruption may create different distributional effects depending on who is allowed to violate the law to obtain economic and legal privileges.