In order to describe quantitatively the highly spatial variability of soil profiles in falluvial soil, based on 139 observed soil profiles in a 15 km2 region in North China Plain, indicator variogram and transiogram were used to analyze the spatial variability of five soil textural layers respectively. Results of indicator variograms showed that the soil textural layers in horizontal direction had strong stochastic character whereas the results of transiogram showed that the soil textural layers in horizontal direction had a certain degree of regularity. Indicator variogram model was used in Sequential Indicator Simulation (SIS). Three dimensional training image was created by using isotropy distribution proportion and anisotropy mean lengths obtained from transiogram, and then, the training image was used as input to multiple-point geostatistics simulation (MPS). The simulation results were validated and compared with traditional SIS method. Results showed that the MPS method is better than SIS method both in prediction accuracy and reproduction of connectivity of soil textural layers in vertical direction. The results, therefore, can better reflect the spatial character of soil textural layers.