This paper quantifies the experimentally evaluated life-cycle benefits of a widely implemented early childhood program targeting disadvantaged families. We join experimental data with non-experimental data using economic models to forecast its life-cycle benefits. Our baseline estimate of the internal rate of return (benefit/cost ratio) is 13.7% (7.3). We conduct extensive sensitivity analyses to account for model estimation error, forecasting error, and judgments made about the empirical magnitudes of non-market benefits. We examine the performance of widely used, ad hoc estimates of long-term benefit/cost ratios based on short-term measures of childhood test scores and find them wanting. (author abstract)