Through the evaluation of the soil nutrient in a 7-year long-term fertilization experiment of wheat-maize rotation with different phosphorus application levels, the inorganic phosphorus fractions under the phosphorus depletion or accumulation in calcareous fluvo-aquic soils were studied. Path analysis and multiple regression analysis were performed to investigate the relationships among various soil available phosphorus determined by three methods, so did phosphorus fractions. The results showed that: (1) Compared with the initial soil, the total phosphorus under N0P0K0 and N2P0K2 were decreased by 15.2% and 29.7%, respectively, and the total inorganic phosphorus decreased by 13.5% and 11.8%. The total phosphorus under N2P2K2 and N2P3K2 were increased by 8.2% and 27.2% respectively, while the total inorganic phosphorus increased by 11.1% and 27.8%. The formations of inorganic phosphorus in the tested soil were mainly Ca₁₀-P and Ca₈-P. The application of phosphorus fertilizer could increase the relative proportions of Ca₂-P, Ca₈-P, Al-P and Fe-P. (2) In the depletion state of phosphorus, the inorganic phosphorus used by plants mainly came from slow-release phosphorus sources (Ca₈-P, Al-P, Fe-P; 75%), un-available phosphorus sources Ca₁₀-P (11.5%~14.0%), and available phosphorus sources Ca₂-P (7.5%~8.9%). At inorganic phosphorus surplus condition, excess inorganic phosphorus mainly converted to Ca₈-P (50%~70%), Al/Fe-P (10%~23%), O-P (8%), and Ca₂-P (0.2%~1.8%). (3) Ca₂-P and Al-P had positive effects on available phosphorus extracted by three methods with the greater contributions. Ca₂-P and Ca₈-P could be effectively extracted by the Olsen method. While, the Mehlich3 method mainly extracted Ca₂-P, Ca₈-P and Al-P, and the main extraction of AEM were soil Ca₂-P and Fe-P. (4) All the three methods were feasible for determining soil available phosphorus contents, and the Olsen method was the best.