Observation of soil salt content plays an important foundation role in enhancing the research of monitoring and prediction for soil salinity. In order to study the problem of soil salinity in the oases of arid regions，we use traditional statistics and geostatistics combined with the geographical information system(GIS) technology to investigate the spatial variability and spatial distribution of soil salt content in the delta of the Weigan and Kuqa rivers. The results show that the soil salt content in three successive layers in this oasis exhibit moderate spatial variation, and the degree of this variation decreases as soil depth increases. Both the structure and the stochastic factors are of important influence for the spatial variation, and have a strong spatial autocorrelation. The rational sampling quantity of three layers should be at 0.05 level with a 20% relative error. Gaussian models were fitted for semi-variance of ＞10-30cm and ＞30-50cm layers. Considering the compound scale effect of spatial variation, the nested spherical models were fitted for semi-variance of topsoil. The kriging model and the isogram maps accurately demonstrate the spatial distribution and the variation tendency. These research results have provided a reference for the survey and quality evaluation of soil salt content and can be used as a guide for meliorating and utilizing the soil salinity.