To relate short-term nutrient balances with long-term changes in soil fertility, we compared soil availablenutrient stores with fertilizer application, crop productivity, and nutrient balances. Stores of nitrogen (N: quantity of available N), phosphorus (P: quantity of Truog-P), and potassium (K: quantity of exchangeable K) in the topsoil were calculated five times from 1979 to 2003. In each survey, nutrient inputs by fertilizer, manure, and soil amendment and output by crop removal were calculated. The arithmetic average was calculated for seven crop groups [(paddy rice, Oryza sativa L.), upland crop (wheat, Poaceae triticum L., etc.), vegetables (radish, Raphanus sativus L. etc.), orchards (orange, Malus pumila Mill. etc.), tea (Camelia sinensis L.), forage (Dent corn, Zea mays L. etc.), fodder (timothy, Pheleum pratense L., etc.)] on 16 groups of soils. In spite of large changes in the N balance and in manure N input, available N was relatively constant during the five surveys both within and between crops. On the other hand, the available P and K stores and balances changed greatly during the surveys. The quantities of N, P, and K applied in manure had no consistent effect on the soil available N, P, and K stores. The Truog-P store exceeded guidelines in paddy rice, orchards, and tea. The K store exceeded guidelines in upland crops, vegetables, orchards, and tea. Although the available N store was managed well, the available P and K stores were not. Sustainable agriculture that makes effective use of nutrients and has a low environmental impact requires the use of both nutrient balances and available nutrient stores.