The key to application of the Indicator Kriging method is the choice of an appropriate threshold. Three soil salinity thresholds, i.e. 1.0 g kg^-1, 2.0 g kg^-1 and 3.0 g kg^-1 were set for the study on variograms, prediction probability and spatial distribution of the prediction probability of soil salinity in the top layer (0~20 cm) of a farmland in relation to soil salinity threshold in Yucheng, typical of the salt-affected soil amelioration zone in the Huang-Huai-Hai Plain. Results show（1）regardless of threshold values, salt contents in all soils are all moderate in spatial autocorrelation; however, with threshold declining, soil salinity intensifies in spatial structure and its model of variation function increases in accuracy; therefore, taking into account the accuracy of the indicator variogram model, the soil salt content of 1.0 g kg^-1 is the optimal threshold for assessment of the risk of salinization of the studied region; （2）both the maximum value and the mean value of prediction probability increase with rising salinity threshold, which may serve as reference for selecting thresholds for different objects of soil salinization risk assessment; and（3）predictive distribution of probability prediction of soil salinity using different thresholds demonstrates certain spatial regularity and similarity; high probability is concentrated mainly in the west part of the studied region, while low probability in the east. Distribution of the prediction probability of soil salt content is closely related to landform and topography and distribution of rivers in the study area.