Taking into account the problem of soil salinization existing in the current Huanghuaihai Plain, spatial distribution of soil salinity at depth of 0 – 20 cm layer at different scales was analyzed in a typical country of Yucheng City with the application of GIS and indicator Kriging of nonparametric geostatistics, and probability maps of soil salinity were provided. The results indicated that neither soil salinity of country scale nor of town scale followed normal distribution because of the existence of outliers, but robust semivariogram could be obtained by using indicator Kriging. Soil salinity of the both scales exhibited moderate spatial autocorrelation due to compound impacts of structural and stochastic factors. The C0/Sill decreased, but range increased as the enlargement of research-scale, which indicate that the self-correlation of soil salinity increased, the influences of random factors weakened while these of structural factors increased. Salinization risk increased with the decrease of scale.