Nutrient intakes estimated using a 24-hr recall, a dirt record and a food frequency questionnaire(FFQ) were compared in a group of ninety-four elderly people(21 males, 73 females) in Cheongju, a city in Chung-Buk province. Mean intakes for energy, protein, Ca, p, Na, K thiamin and niacin obtained from the diet record were higher than those from the 24-hr recall. Mean intakes for energy, protein, Ca, P, Na, K thiamin, niacin and vitamin C from the FFQ were higher than those obtained from the 24-hr recall or the diet record. Correlation coefficients between the nutrient intake values from the 24-hr recall and those from the diet record ranged from 0.84 to 0.95 and were significantly correlated(p < 0.001). About 80% of the subjects in the lowest quintile by the 24-hr recall were also in the lowest two quintiles by the diet record. While the percentage fallen into the opposite category ranged from 0% to 15%. For most nutrients, at least 65% of the subjects when classified by the 24-hr recall fell into the same quintile when classified by the diet record, and the mean kappa value was 0.7. About 52% of the subjects in the lowest category by the 24-hr recall fell into the lowest two categories by FFQ. The mean percentage of the subjects in the lowest quintile by the diet record or in the lowest two quintiles by the FFQ was 51%. For most nutrients, at least 24% of each of the subjects when classified by both the 24-hr recall and the diet record fell into the same category when classified by the FFQ. The kappa values between the 24-hr recall or the diet recall and the FFQ were 0.17. These data indicate that in elderly subjects the 24-hr recall can provide very similar information to that obtained from the diet record but the FFQ can not provide good information if the right FFQ method is not used for only elderly subjects.