OBJECTIVES: Childhood sociodemographic, psychosocial, and environmental factors are often assumed to affect adult health and longevity. These relationships were prospectively tested by using the 7-decade Terman Life Cycle Study of Children With High Ability (n = 1285). METHODS: Parental socioeconomic status, childhood health, objective childhood stressors (e.g., death or divorce of parents), and childhood personality were considered as potential predictors in hazard regression analyses of longevity through 1991. RESULTS: Parental divorce during childhood predicted decreased longevity, with sex controlled. Other potential social predictors failed to show significant associations with longevity. Three dimensions of childhood personality--conscientiousness, lack of cheerfulness, and permanency of mood (males only)--predicted increased longevity. The effects of parental divorce and childhood personality were largely independent and did not account for any of the gender difference in mortality. CONCLUSIONS: A small number of childhood factors significantly predicted mortality across the life span in this sample. Further research should focus on how these psychosocial factors influence longevity.