The spatial change of textural layers is one of the important features of alluvial soils. It affects soil water transport and solute movement in field intensely. But the quantitative studies on it are limited. Based on the results obtained in our research,this paper presented two stochastic models- MC (Markov chain) model and MC-LN (Markov chain Lognormal distribution) model for simulating textural profiles of regional alluvial soils using Markov chain theory, and simulated soil textural profiles by Monte Carlo method. Results showed that the vertical changes of textural layers along soil profiles in the research region could be reflected by this two models.But in the probability distribution of textural layer thickness, MC model often brought on more error, and MC-LN model was in accord with the measured data.Considering the far more intensive effects of number and sequence of textural layers in profiles than the error of textural layer thickness on field water and solute movement. two models would be accepted simultaneously.