The role of dietary fat and cholesterol in the low blood cholesterol-cancer association was examined in the National Health and Nutrition Survey Epidemiologic Follow-up Study (NHEFS) conducted during 1971-1984. Of the 9,593 male and female participants, aged 25-74 years, 638 developed cancer during the median of 10.2 years of follow-up. Diet was assessed using a 24-hour dietary recall. Despite an inverse association between serum cholesterol and cancer, male cases consumed, on average, more dietary cholesterol, a larger proportion of calories as saturated fat, but a similar proportion of calories as linoleic acid. Among females, cases consumed less fat and cholesterol than noncases. Among males with central adiposity (subscapular-to-triceps skinfold ratio greater than or equal to 1), cases ate more dietary cholesterol both absolutely and as a proportion of calories. Among males with peripheral adiposity, cases tended to eat less dietary cholesterol than noncases. Among females, serum cholesterol was inversely associated with cancer risk among those with central adiposity but directly associated among those with peripheral adiposity. Dietary comparisons were adjusted for age, race, body mass index, smoking status, alcohol consumption, education, and economic status. The results indicate that diet cannot explain the low blood cholesterol-cancer association in men. Furthermore, inconsistencies in the literature in this area may be due, in part, to differing proportions of central adiposity in study populations.