[Background] Topographic factor (LS factor) including slope length (L factor in variable λ) and slope gradient (S factor in variable θ) is one essential parameter of Revised Universal Soil Loss Equation (RUSLE). LS factor is greatly influenced by Digital Elevation Model (DEM) resolution. However, most researchers adopted only one specific DEM resolution, and the influences caused by DEM resolution were commonly ignored. The objective of this paper was to explore the impacts of DEM resolution on LS and provide certain experience for DEM resolution correction in a specific scale. [Methods] The black soil region,loess region, and red soil region were chosen as the study areas. At first, we re-sampled the ASTER GDEM (30 m) to 6 other DEM esolutions,namely,60 m,90 m,120 m, 250 m,500 m and 1 000 m using the platform of ArcGIS Desktop 10.0. Then, the LS factor in these regions was calculated using Arc Macro Language (AML) in the platform of ArcInfo Workstation 9. x. Finally, quantitative relation between LS and DEM under different resolutions was explored for studying the effects of DEM resolution on the accuracy of LS. [Results] LS factor in black soil region was the minimum, and that in the red soil region was the maximum. The S factor was more sensitive to the variation of DEM resolution compared to the L factor. With the DEM resolution decreasing,the overall change trend of L factor was not very significant. Whilec S factor and the standard deviation of LS factor gradually decreased. When the grid size of the DEM increased,all LS factor decreased in all of the three soil regions with different magnitude. With the DEM resolution reduced from 30 m to 1 000 m,LS factor dropped to 1/6 - 1/3 of that in the resolution of 30 m. When the DEM resolution was 30 m, the correlation coefficient between LS factor and slope degree, and that between LS factor and the slope length was 0.78 - 0.86 ( P < 0.05), and 0.34 - 0.46 ( P < 0.05), respectively. With the decreasing of the DEM resolution, correlation coefficient tended to be significantly lower. In terms of its spatial distribution,LS factor value tended to be much larger in the middle part of steep slopes comparing with the peak and foot of the mountain. When the DEM resolution was 1 000 m,LS factor value in more than 60% of the whole study area was at the lowest class. [Conclusions] The impact of DEM resolution on the LS factor is significant and it should not be ignored when using distributed soil loss model. We should take necessary measures to reduce the impact of DEM resolution on the LS factor,and the results of this paper may provide certain references for DEM resolution and correction parameter selection in large and medium scale.