In order to improve the drawing accuracy of soil salt distribution map or reduce the cost of field sampling and indoor analysis，it is very necessary to optimize the layout of soil sampling points. Taking the minimum regression Kriging estimated variance (RKV) and the minimum ordinary Kriging estimated variance (OKV) as the objective function，the layout of soil salt sampling points was optimized by spatial simulated annealing algorithm (SSA)，the selection of covariates in regression Kriging (RK) method was closely related to soil salt distribution and easy to obtain environmental variables. The optimization results were evaluated by the nearest distance distribution function (G function)，interval distance distribution function (F function) and K-S test. The results showed that wasteland distribution on a large scale (400m×400m) was significantly correlated with salt，but weak correlated on a small scale (50m×50m).Taking the distribution of wasteland (the shortest distance from wasteland) as the collaborative variable and optimizing the sampling layout by RK method (sampling number was 55)，the mean RKV of the study area was 0.0807，which was 29.39% smaller than the mean OKV(0.1143). The G function curve and F function curve of the optimized sampling set all fall within the 95% confidence interval，there was no significant difference between the F function and G function and their theoretical values, the layout of the sampling points was well arranged and covered the geographical space. K-S test showed that the distribution of the shortest distance from wasteland on the optimized sampling point and its population obeyed the same probability distribution，the layout of the sampling points was well arranged and covered the attributes space. The research results can provide reference for the layout of soil salinization monitoring points in Hetao Irrigation Area.