The closure of low-performing schools is an essential feature of the charter school model. Our regression discontinuity analysis uses an exogenous source of variation in school closure an Ohio law that requires charter schools to close if they fail to meet a specific performance standard to estimate the causal effect of closure on student achievement. The results indicate that closing low-performing charter schools eventually yields achievement gains of around 0.2-0.3 standard deviations in reading and math for students attending these schools at the time they were identified for closure. The study also employs mandatory closure as an instrument for estimating the impact of exiting low-quality charter schools, thus providing plausible lower-bound estimates of charter school effectiveness. These results complement the more common lottery-based estimates of charter school effects, which likely serve as upper-bound estimates due to their focus on oversubscribed schools often located in cities with high-performing charter sectors. We discuss the implications for research and policy. (C) 2016 Elsevier Inc. All rights reserved.