Contributions of specific foods to absolute intake and between-person variance in 13 nutrients were examined to develop a short food frequency questionnaire (FFQ) for the Korean elderly using dietary data collected by a semiquantitative FFQ with 96 food items. The contributions of specific foods to 90% absolute nutrient intake were measured by assessing their percentage to the total consumption of a nutrient by our samples(n=98). To assess the contribution of the foods to the between-person variance in intake of each specific nutrient, stepwise multiple regression was performed. Cumulative R2 served as the measure of contribution to variation in intake. The most important food source of absolute intake was rice for energy and protein. Important foods of between person variance contribution include bread for energy and pork for protein. Important foods of absolute and variance contribution of nutrient intake were similar for fat(noodles including jajangmyun and kalguksu), calcium(anchovy and milk) and vitamin A (carrot). The number of foods necessary to account for the variation in nutrient intake among older people in Korea differ considerably by nutrient: 18 - 43 and 8-26 foods were accounted for the respective 90 and 80% of absolute intake depending on nutrients. Carbohydrates and vitamin A and C had relatively a few major sources, thus may be assessed well by a small number of foods. Conversely, iron, potassium and B vitamins appear to need longer lists of foods. Fewer numbeer (6 - 12 and 4 - 9) of foods were required for the corresponding percentages of between-person variation for all nutrients. Depending on nutrients, 0.980-0.995 and 0.924-0.987 were respective correlation coefficients of nutrient intakes measured by 96 food items with those by lists of foods for 90% and 80% of absolute intake contribution. These findings suggest that an incomplete food list on the basis of its contribution to absolute and between person variation intakes may still be effective in discriminating among individuals. T