The cross-sectional correlates of three hemostatic factors--fibrinogen, factor VII, and factor VIII--were examined in the Cardiovascular Health Study, a population-based cohort study of 5,201 subjects over age 65 years. Subjects were recruited in 1989-1990 in Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Pittsburgh, Pennsylvania. In multivariate linear regression models, cardiac risk factors significantly associated with fibrinogen were current smoking, race, lipids, and white blood count. In women, alcohol use, obesity, physical activity, and insulin level were also significant, while in men hypertension was correlated. The significant correlates of factor VII were lipids and white blood count in men and estrogen use, alcohol use, race, lipids, insulin level, white blood count, and obesity in women. The independent correlates of factor VIII were insulin, glucose, and race in both sexes; low density lipoprotein cholesterol, white blood count, and diuretic use in men; and alcohol use in women. In multivariate models, factors known to be modifiable risk factors for cardiovascular disease accounted for more of the population variance of these hemostatic factors in women than in men, especially for factor VII. The hemostatic factors may mediate some effects of risk factors on disease, and this should be considered in longitudinal studies.