It is difficult to measure the contribution of school quality to student outcomes. Since parents choose where their children attend school, it is impossible to tell using a naïve analysis whether students at 'good' schools do well because the schools are good or whether they do well because they are in some sense 'good' students. For instance, if all the best students go to school A rather than to school B, then school A will appear to be better than school B. This is true if school A is indeed better or counter-intuitively even if it is slightly worse. To distinguish the quality of the school from the quality of the students, it is necessary to understand how parents choose different schools for their children and which students end up at different schools. Our current research is using a model of parental choice of schools to analyse data on educational choices and educational outcomes in the United States using the National Educational Longitudinal Survey 1988 (NELS), a dataset that has more detailed information on schools and their students than comparable datasets available in the United Kingdom. Two papers are nearing completion. These papers were presented at the May 27th European Commission workshop on Quality and Efficiency in Education and Training, at the June 17-20 North American Summer Meeting of the Econometric Society, and at the August 20th-24th European Meeting of the Econometric Society.