In epidemiological studies of diet and chronic disease, a brief yet comprehensive diet history questionnaire must aggregate some foods into food groups. A nutrient density is assigned to each food group by averaging the densities of its constituent foods. A person's intake of a given nutrient is then estimated by multiplying the reported consumption of each food group by its average nutrient density and summing over food groups. These calculations could introduce bias in multiethnic studies, if the average nutrient densities assigned to food groups are inappropriate for some ethnic populations. This issue is examined here for intakes of total fat, saturated fat, and vitamin A for U.S. blacks and whites. We used 24-hour diet recall data from the Second National Health and Nutrition Examination Survey (NHANES II) to assess black-white differences in relative frequency of consumption of foods within food groups of a diet history questionnaire. We also calculated ethnic-specific average nutrient densities for each food group by weighting the densities of its foods in proportion to their frequency of consumption by black and white NHANES II participants. We found black-white differences in the frequency of consumption of foods within 14 food groups. However, blacks and whites had different average total fat densities for only 1 of the 14 food groups, no difference in saturated fat densities for any food group, and different vitamin A densities for 2 food groups. Among blacks and whites, there is no advantage to calculating ethnic-specific average nutrient densities for food groups comprised of foods with similar densities.(ABSTRACT TRUNCATED AT 250 WORDS)