It is generally accepted that early childhood education improves the cognitive performance of children in poverty in the short-term, but whether cognitive effects persist in the long-term is hotly debated. This paper presents the results of a critical review of 38 studies of the long-term effects of early childhood programs on children in poverty. Outcomes examined include IQ, achievement, and academic success as measured by grade repetition, special education placement, and high school graduation. Early childhood education is found to produce persistent effects on achievement and academic success, but not on IQ (with some exceptions). Head Start and public school programs produce the same types of effects as better funded model programs, but at least some of the effects are smaller. Cost-benefit analysis based on one randomized trial finds that the economic return from providing early education to children in poverty far exceeds the costs. Head Start, public school preschool education, and education in high-quality child care programs all offer avenues for government investment to improve the long-term cognitive development and academic success of children in poverty.