This paper explores the electoral effects of providing information on the educational outcomes of municipal schools when the mayor is running for reelection. We designed and implemented an experiment in Chile whereby we sent 128,033 letters to voters in 400 randomly selected polling stations prior to the 2016 municipal elections. The letters included information on past test scores for local public schools (levels and changes), and either average or maximum outcomes for comparable municipalities. Our findings do not reveal a relevant average impact of the letters, but when they contain poor educational outcomes, voter turnout decreases, translating almost one to one in decreases in votes for the incumbent mayor. Voters respond to educational results in levels and to letters that have average results as a benchmark. The results are especially strong when poor educational outcomes come as bad news to voters. We also find spillover effects in the municipal council election. Overall, our findings suggest that voters hold politicians accountable when faced to certain (but not all) types of information on their performance. JEL Codes: D72, H75, I25.