This article examines the effects of school vouchers on student test scores in New York, New York, Dayton, Ohio, and Washington, DC. The evaluations in all three cities are designed as randomized field trials. The findings, therefore, are not confounded by the self-selection problems that pervade most observational data. After 2 years, African Americans who switched from public to private school gained, relative to their public-school peers, an average of 6.3 National Percentile Ranking points in the three cities on the Iowa Test of Basic Skills. The gains by city were 4.2 points in New York, 6.5 points in Dayton, and 9.2 points in Washington. Effects for African Americans are statistically significant in all three cities. In no city are statistically significant effects observed for other ethnic groups, after either 1 or 2 years. © 2002 by the Association for Policy Analysis and Management.