[Objective] To test the rationality of soil sampling design, we took Sanyanjing basin of Shouyang county in Shanxi Province as a case study, to analyze the uniformity and representativeness of soil water sampling.[Methods] Using statistics analysis software and ArcGIS, we took grid sampling points as the initial layout. And then combined with remote sensing images, the automatic generated regular grid sampling points were finely adjusted giving the consideration of land use types and operation, namely, using uniform sampling in large scale and random sampling in small scale method for soil sampling point layout.[Results] Both of the nearest neighbor index and coefficient of variation showed a uniform distribution of sampling points. Sampling points obtained by uniform sampling in large scale and random sampling in small scale method were mainly concentrated in the slope of 0°~15°, which is not representative according to the distribution of slope. While the selected sampling points were representative according to the distribution of aspect. The points obtained by uniform sampling in large scale and random sampling in small scale method were more representative in grassland and farmland, but fewer representatives in forest land, due to the woodland is mainly distributed on steep slope and difficult to sample.[Conclusion] The uniformity and representativeness of sampling points obtained by large uniform small sample random sampling method are quite good. This sampling method can be well applied in hilly and gully area of the Loess Plateau to obtain uniform and representative sampling points.