Using a large administrative data set on all state schools in England, this paper studies the effect of free part-time preschool education at age 3 on child outcomes in primary school at ages 5, 7 and 11. To do this it exploits the staggered implementation of free preschool places across Local Education Authorities in England. We demonstrate that the policy led to a substantial transfer to parents as 3 of every 4 places funded were already being paid for privately, while only one genuinely new place was created. Despite this large crowd-out of parental investments the policy had some small beneficial effects at age 5 with a 10pp increase in the proportion of 3-year-olds covered by free places improving individual school outcomes by around 2% of a standard deviation. Effects are somewhat larger for boys and for more disadvantaged children. Impacts are twice as large for children in LEAs where more new places were created, which implies that benefits came from additional participation, not from income effects. Effects of the policy at age 7 are very small, with no benefits at age 11.