Recent research suggests that childhood cognitive performance is associated with various health outcomes, but the nature of the relation is not well understood. It is unclear whether the association occurs across the continuum of cognitive performance, and if it is independent of socioeconomic status. Prospective data from the Terman Life Cycle Study were used to evaluate the hypothesis of a monotonic relation between childhood intelligence quotient (IQ) and adult mortality and to determine whether there exists a threshold beyond which the protective effects of IQ are no longer evident. A total of 897 individuals of school age who scored 135 or higher on the Stanford-Binet IQ test were recruited in 1922. Cox proportional hazards regression was used to estimate survival over a 64-year period. A 15-point advantage in childhood IQ was significantly associated with a decreased risk of mortality (hazard ratio = 0.68, 95% confidence interval: 0.49, 0.93) for IQ scores up to 163; beyond that, the risk of death plateaued. Results were similar when the sample was limited to those participants whose fathers had nonmanual occupations. Childhood IQ, even at the upper end of the distribution, is a significant predictor of mortality, independent of childhood social position.