Research on the effect of income inequality on pollution shows mixed results. This paper takes a new look at the urban air pollution data set of the U.N.'s Global Environmental Monitoring System (GEMS). We investigate the impact of income inequality on urban air pollution and relate the results to a median-voter model. In this model, more income inequality decreases the median income, reduces pollution controls, and increases output and pollution when the median income is above a threshold. We find that income inequality, measured by the Gini coefficient, increases SO2 concentration in the rich democracies. The estimated effects are non-negligible in size. For the poor non-democracies, we find no evidence that the Gini coefficient impacts SO2 concentration. The Gini coefficient is estimated to increase the smoke concentration at the five- or ten-percent significance level in most specifications in a pooled sample. We find no evidence that the Gini coefficient impacts the concentration of particulates in a pooled sample. We conclude that the empirical results largely are consistent with the medianvoter model.