[Objective] Soil erodibility is a key factor of calculating the soil erosion, and the investigation of cultivated soil erodibility provide the scientific basic to calculate and evaluate soil erosion in Shaanxi Province.[Methods] The soil data was from nine experimental plots that distributed in Shaanxi Province of China. To calculate soil erodibility of these areas, we utilized five commonly used models, which are universal soil loss equation(USLE), revised universal soil loss equation version 2(RUSLE2), Erosion productivity impact calculator(EPIC), geometric mean diameter(Dg) and geometric mean diameter-organic matter(Dg-OM).[Results] The equation of very fine sand(VFS) in RUSLE2 underestimated 14.53% of VFS content in hilly area of Losses Plateau of Shaanxi Province and overestimated 32.91% of VFS content in the Southern Shaanxi Province. Based on the measured values, the revised equation reduced the average calculation error of VFS content to 7.81% and 13.14% respectively. Secondly, comparison of K values, the measured K value in Zizhou County is 0.002 69[(t·hm²·h)/(hm²·MJ·mm)], the mean K value of Dg-OM model is 0.029 7[(t·hm²·h)/(hm²·MJ·mm)]. Thirdly, the interrill erodibility(K_i) and rill erodibility(K_r) parameters in water erosion prediction project model(WEPP) were calculated and the correlation of erodibility parameter between K_i, K_r and K of USLE were 0.738 6 and 0.607 4.[Conclusion] The average calculation error of revised equation of very fine sand is less than the equation of very fine sand in RUSLE2. Dg-OM model was suitable for calculation the soil erodibility in hilly area of Losses Plateau of Shaanxi Province and the area of Changwu County, Yangling District and Ankang City. Moreover, K_i and K_r parameters in WEPP was found the good correlation of erodibility parameter between USLE and WEPP model when the sand content was less than 30%.