Although several risk factors for heart disease including high blood pressure, diabetes mellitus, and lipid and lipoprotein abnormalities are associated with overweight, overweight is not consistently associated with coronary heart disease risk. Some prospective studies of white men (life insurance cohorts, airline pilots, cancer study volunteers, and the Framingham population) have shown a positive linear relationship of weight to coronary heart disease. Other epidemiologic studies show a negative association, no association, a U-shaped relationship, or a threshold effect. The inconsistencies do not appear to be explained by differences in the definition or distribution of obesity, duration of follow-up, or risk factor distribution. Neither misclassification bias nor confounding by cigarette smoking or chronic disease appears to explain the inconsistencies. No known protective effect of obesity could explain these divergent findings. Inconsistent results with regard to the nature, strength, and linearity of the association between obesity and atherosclerosis do not support the hypothesis that obesity causes atherosclerosis, despite its biological plausibility.