Excessive application of animal manure composts engenders soil phosphorus accumulation. Generally, animal manure composts have lower Nitrogen-to-Phosphorus ratios (N/P ratio) than those of crop uptake. When animal manure composts were applied based on Nitrogen (N) amounts that meet crop requirements, the associated amounts of Phosphorus (P) applied were greater than the crop demand. Consequently, excess P accumulates in the soil. The accumulated P can cause water system eutrophication through surface runoff and soil erosion (Sharpley, 1999). An application system of animal manure compost without increasing soil P levels is needed to establish environmentally conscious agriculture. Eghball and Power (1999) reported that annual or biennial beef cattle manure applicatio or compost application based on N or P requirements of corn engendered corn yields that were similar to those for fertilizer application. Annual P-based manure or compost treatments had similar available soil P levels to the original soil P level before treatment application. The effectiveness of a P-based application system of animal manure composts has not been reported for paddy fields. In this study, we applied cattle and poultry manure composts with N-based and P-based systems to paddy fields and examined effects of two application systems of animal manure compost on rice yield and P-accumulation in paddy soil. Field experiments using rice (Oryza sativa L. cv. Hitomebore) were conducted in 2004 and 2005 at the Field Science Center, Tohoku University, Japan. The soil was classified as Melanic Andosol according to WRB (1998). Available N contents of the composts were estimated from total N content and the nitrogen mineralization rate (about 30%). Available P contents of the composts were determined from the sum of water extractable and 0.5 M sodium bicarbonate extractable phosphorus, according to the results of Ito et al. (2004). Treatments were N-based cattle manure compost application (N-based CC), P-based cattle manure compost application (P-based CC), N-based poultry manure compost application (N-based PC), P-based poultry manure compost application (P-based PC), Chemical Fertilizer application (CF), and no compost and fertilizer application (CK). The standard application rate was N-P₂O₅-K₂O=6-6-6 g m^-2. Compost application rates were determined to provide 6 g m^-2 of available N in N-based treatment, and 6 g m^-2 of available P₂O₅ in P-based treatment. In P-based treatment, nitrogen fertilizer (mixture of polyolefin coated urea with a dissolution rate of 70 and 30 days (2:1 w/w)) was applied respectively at rates of 3.3 and 3.5 g m^-2 of N in P-based CC and P-based PC in 2004, and respectively at rates of 3.1 and 3.8 g m^-2 of N in P-based CC and P-based PC in 2005. These six treatments were arranged with randomized block design with three replicates. Brown rice yields were determined, along with nitrogen and phosphorus uptake, and soil test P assessed by modified Bray 2 method with an extractant to soil ratio of 20. Brown rice yields (g m^-2) ranged from 506 to 710; the order was CF≥P-based PC≥N-based PC≥P-based CC≥N-based CC≥CK in 2004. In 2005, rice yields ranged from 394 to 629; the order was P-based PC≥CF≥N-based PC≥P-based CC>N-based CC>CK. The yields in N-based and P-based PC treatments were equal to or greater than those of the CF treatment, and the yields in N-based and P-based CC treatments were lower than that in CF. No significant (P<0.05) differences in yields were found between five treatments except CK treatment in 2004, and the yields in N-based CC treatment were significantly lower than those in CF treatment in2005. The reason for lower rice yields in cattle manure compost treatments was considered that the nitrogen mineralization rate of cattle manure compost was lower than the assumed value. Soil test P values were significantly increased by N-based PC treatment over those of CK, when soil test P was determined with plow layer soils collected at harvest times in both of 2004 and 2005. The N-based and P-based CC, P-based PC and CF treatments did not significantly increase the soil test P over that of CK soils after rice cultivation in either 2004 or 2005. The P-based PC treatment suppressed the soil test P increase to 9% and 33% against those in N-based PC treatments in 2004 and 2005, respectively. Soil test P values showed a close exponential relationship with total P inputs by composts and chemical fertilizer in 2005. From the above results obtained from two years ' research, phosphorus-based poultry manure compost application was inferred to maintain rice yields equal to or greater than conventional chemical fertilizer systems; it was also inferred to suppress soil phosphorus accumulation in Andosol paddy fields.