Four measurements of per-capita education expenditure are used to analyze its effect on income inequality at the county-level in Chile. The latter is quantified through the Gini coefficient, Theil index, and the S80/S20 and S90/S10 ratios. The longitudinal study considers 316 Chilean counties in order to estimate a fractional response probit model for the Gini coefficient and random-effects Tobit models for the remaining indicators of inequality. Main findings show that an increasing per-capita education expenditure would reduce the income inequality measured by Theil index, but it would accentuate the gap between the poorest and wealthiest decile/quantile. And a greater education expenditure on personnel and teaching staff would enlarge the Gini coefficient and S90/S10 ratio. In addition, income inequality at the county-level is exacerbated by a larger indigenous and rural population. Finally, the evidence reveals that a convex Kuznets-curve exists for the extreme values of income distribution.