Schools across the United States are facing budgetary pressures on a scale not seen in generations. Times of fiscal exigency force policymakers and education practitioners to pay more attention to the return on various categories of public investment in education. The sizes of the classes in which students are educated are often a focus of these discussions because they are a key determinant of educational spending. The declines in funding currently faced by many schools mean that cuts must be made, but it is often unclear how to make cuts in ways that minimize harm to students. This article reviews the evidence base available to inform such policy decisions. It divides the review of the high-quality evidence on class size into three sections. First, it discusses the Tennessee STAR experiment, which is the most important and influential study because it is the only modern randomized experiment conducted at a significant scale. Second, it reviews the quasi-experimental evidence based on naturally occurring variation in class size that is credibly exogenous to student achievement. Finally, it reviews the quasi-experimental evaluations of two statewide class-size reduction policies. These studies are examined separately because in addition to offering evidence about class size, they also raise important issues related to the design and implementation of class-size policies. (Contains 2 figures, 1 table, and 24 footnotes.)