Residential mobility is likely to have consequences for student performance, but prior empirical work is largely correlational and offers little insight into its impacts. Using rich, longitudinal data, we estimate the effects of residential mobility on the performance of New York City public school students. Using both student fixed effects and instrumental variables approaches, we find that long-distance moves have negative effects, while short-distance moves improve student performance. These differential effects are partially, but not fully, explained by school mobility. Rather, the positive effects of short-distance moves may be explained by improvements in housing, while the negative impacts of long-distance moves may be explained by lower performance relative to school peers and loss of social capital.