Many states have implemented or expanded state-funded prekindergarten programs in the last decade, encouraged by claims about the benefits that can be expected and the importance of early experiences for children's development, especially for economically disadvantaged children. However, there is remarkably little methodologically adequate evidence about the effects of such programs. Using a subsample of children with parental consent from a larger sample of children randomly assigned to attend the Tennessee pre-k program or not, this study examined effects on cognitive and noncognitive outcomes through third grade. At the end of the pre-k year, program participants showed better outcomes than comparable nonparticipants on achievement measures and ratings of school readiness by kindergarten teachers. But those effects were not sustained in subsequent years and, indeed, by the end of third grade the pre-k participants scored lower on the achievement measures than nonparticipants. These results raise questions about the way state pre-k programs have been designed and implemented.