A large-scale experiment is described in which kindergarten students and teachers were randomly assigned to small and large classes within each participating school. Students remained in these classes for 2 years. At the end of each grade they were measured in reading and mathematics by standardized and curriculum-based tests. The results are definitive: (a) a significant benefit accrues to students in reduced-size classes in both subject areas and (b) there is evidence that minority students in particular benefit from the smaller class environment, especially when curriculum-based tests are used as the learning criteria. A longitudinal analysis of a portion of the sample indicated that students in small classes outperform their peers in kindergarten classes of regular size and also gain more in reading outcomes during the second year. The question of why these effects are realized remains largely unanswered, but in light of these findings, is particularly important to pursue.