Dietary consumption data obtained by food weighing for 539 households in Taiwan in 1980-1981 were used to search for predictor foods of individual intake of several vitamins and calcium. Stepwise multiple regression analysis was carried out on food items that were univariately and significantly associated with nutrient intakes, to screen for the important predictors. Less than ten foods were enough to explain more than 90% of the interperson variation for vitamin A, vitamin B1, and vitamin B2. For vitamin C and calcium, 20 food items could explain only 87% of the variation. Salt, brown sugar, and soy bean sauce were selected among calcium predictors. The information on foods predictive of nutrient intake is useful in designing food frequency questionnaires and materials for nutrition education. Although this paper provides such information for Chinese living in Taiwan, some issues described (such as the need to assess spices) may have been overlooked and could be generalized to populations using stir-frying as a major cooking method.