Charter schools have seen dramatic growth over the last decade. However, we know little about how they affect traditional public schools. I look at how charters affect student outcomes in public schools using data from a large urban school district in the southwest. Unlike prior work that relies on school fixed effects, I address the endogenous location of charter schools using an instrumental variables strategy that relies on plausibly exogenous variation in local building supply. Results show that charters induce modest but statistically significant drops in math and language test scores, particularly for elementary students. However, results for middle and high school students show improvements in discipline. (C) 2011 Elsevier B.V. All rights reserved.