To what extent can preschool education improve early learning outcomes and narrow socioeconomic gaps in academic performance in developing countries? This paper exploits within family variation in exposure to preschool due to its gradual expansion across Peru to estimate the effect of two types of preschools on early learning outcomes. I find that having access to a regular preschool improves second grade standardized test scores by 0.09 and 0.1 SD for reading comprehension and mathematics respectively, but find no impact of having access to a community preschool—where a local mother provides the service for students in her community—on test scores. The two preschool modalities are assigned to towns based on the number of preschool-aged students in each town. I exploit discontinuities in this assignment rule to explore the causal impact of being assigned one type of preschool over the other through a regression discontinuity design. I find that for students in towns near the cut-off, being assigned a pre-school with a trained teacher and proper infrastructure has a positive impact on student learning. Finally, I find that having access to preschool is less beneficial for poor students, suggesting that preschool alone is not sufficient for closing socioeconomic gaps in early achievement and complementary measures targeting the poorest students are necessary.