Using data from the Lipid Research Clinics Program for 1972-1983, the study presented here examined weight history and two indices of obesity: the body mass index (BMI) and the triceps skinfold (TSF) thickness. Cox regression analyses with and without adjustment for cardiovascular disease risk factors revealed a significant (p less than 0.05) quadratic association between BMI and all-causes mortality among men, but not women, after an average 8.4 years of follow-up; mortality was relatively high at both extremes of the BMI distribution. The association was stronger among smokers compared with nonsmokers, and it was apparent among male normotensives, but not hypertensives. All TSF-mortality associations and BMI associations with cancer and coronary heart disease mortality were weak and nonsignificant. Among men, per cent weight change in adulthood showed a significant inverse association with all-causes and cancer mortality. Because BMI and weight history were significantly associated with mortality after adjustment for other risk factors, they appear to be independent predictors of mortality among men.