The spatial variability of textural layers is a major factor influencing the field water and solute transport in alluvial soils. To quantify the water transport process at a regional scale accurately, one has to take the spatial variability of textural layers into account. In this paper, a recently presented Markov chain simulation model for soil textural profiles was coupled with a deterministic field water balance model to conduct a stochastic analysis of the field water balance in a 15 km2 alluvial soil region. The aim is to assess the effect of spatial variability of textural layers on the field water balance at a regional scale. By combining simulated soil textural profiles with the field water balance model, the mean values, extreme values, and probability distributions of field water balance variables were calculated. Results showed large differences in the magnitude of soil water balance variables between different profiles. The extreme difference in the water storage in 1 m depth soil during the winter wheat growth varied with time between 175 mm and 180 mm, which accounted for 86.5--135.0% of the mean of the soil water storage. This indicates that the soil water balance variables derived from only a few soil profiles are not representative for the situation of field water balance in the entire region. The simulated root water uptake showed different types of probability density functions when the soil water storage and the deficit of soil water in the field changed with time. The simulated water storage in 1 m depth soil showed obviously a lognormal distribution, but the measured data showed an approximate normal distribution. It may be heterogeneous irrigation and cropping and some other factors that induce this discrepancy. ¿ 2001 American Geophysical Union