We demonstrate the social efficiency of investing in high-quality early childhood education using newly collected data from the HighScope Perry Preschool Project. The data analyzed are the longest follow-up of any randomized early childhood education program. Annual observations of participant outcomes up to midlife allow us to provide a cost-benefit analysis without relying on forecasts. Adult outcomes on the participants' children and siblings allow us to quantify spillover benefits. The program generates a benefit-cost ratio of 6.0 (p-value = 0.03). Spillover benefits increase this ratio to 7.5 (p-value = 0.00).