This paper organizes and synthesizes the literature on early childhood education and childcare. In it, we go beyond meta-analysis and reanalyze primary data sources in a common framework. We consider the evidence from means-tested demonstration programs, large-scale means-tested programs and universal programs without means testing. We discuss which programs are effective and whether, and for which populations, these programs should be subsidized by governments. The evidence from high-quality demonstration programs targeted toward disadvantaged children shows beneficial effects. Returns exceed costs, even accounting for the deadweight loss of collecting taxes. When proper policy counterfactuals are constructed, Head Start has beneficial effects on disadvantaged children compared to home alternatives. Universal programs benefit disadvantaged children.