Abstract
This paper marshals available evidence from both the U.S. and other countries on the effects of private schools, peer effects, and competition to demonstrate that that any gains in overall student achievement from a large scale voucher program are at best likely to be small. Moreover, given the tendency of parents to judge schools in part by the characteristics of a school's students, a universal voucher system would undoubtedly harm large numbers of disadvantaged students. Although the case for a small means tested voucher program is somewhat stronger, it will do little to improve education for low-performing students.