We combine value-added estimates of school quality with preference data from the New York City public schools match to estimate a model of parental demand for school characteristics. Our approach measures the relative weights parents put on: (1) peer characteristics, (2) test score levels,(3) school quality as measured by value-added, and (4) distance to school. We allow for heterogeneity in preferences and school effectiveness, which enables us to determine whether subgroups trade off school characteristics and school quality differently. Finally, we test for comparative advantage in school choice by investigating the extent to which preferences are linked to idiosyncratic components of achievement gains.