This review paper, prepared for the forthcoming Russell Sage volume Changing Poverty, considers the ability of different education policies to improve the learning outcomes of low-income children in America. Disagreements on this question stem in part from different beliefs about the problems with our nation's public schools. In our view there is some empirical support for each of the general concerns that have been raised about public schools serving high-poverty student populations, including: the need for more funding for those school inputs where additional spending is likely to pass a benefit-cost test; limited capacity of many schools to substantially improve student learning by improving the quality of instruction on their own; and the need for improved incentives for both teachers and students, and for additional operational flexibility. Evidence suggests that the most productive changes to existing education policies are likely to come from increased investments in early childhood education for poor children, improving the design of the federal No Child Left Behind accountability system, providing educators with incentives to adopt practices with a compelling research base while expanding efforts to develop and identify effective instructional regimes, and continued support and evaluation of a variety of public school choice options.