Governments are increasingly publishing information about the performance of the services they provide, in part to help citizens hold their elected representatives accountable for government service outcomes. Yet there is little evidence concerning the influence of information about government service performance on voter behavior. This paper examines the degree to which school performance information affects incumbent support in school board elections. A regression discontinuity analysis indicates that voters reward incumbents when local schools achieve the highest available measure of school performance. Voters do not respond to information differentiating between schools in the middle of the performance distribution, however, or to information from a source that lacks credibility. The following are appended: (1) Regression Discontinuity Specification; (2) Cross-Validation Procedure to Select Bandwidths; (3) Tables and Figures. [This research has been supported by a Harvard University grant from the Multidisciplinary Program in Inequality and Social Policy.]