Abstract: Spatial sampling optimization for soil variables must reconcile the conflicts between the acuracy，survey budget，amount and spatial pattern of sampling points，a classic NP-Hard problem.Numerous studies prove that priori information and multi-objective optimization techniques can im-prove sampling efficiency and accuracy. We evaluated the effects of priori information and spatialstratification technique，and mapped the relationship between the particle and soil sampling solution，and proposed a spatial sampling optimization model for soil variables on the basis of a knowledge-in-formed multi-objective particle swarm optimization algorithm. The model combines minimum Krigingvariance and maximum entropy as the objectives，and optimal sampling size，spatial interval betweensampling sites and spatial accessibility as the sampling constraints，and employs goal programmingmethod to solve the multi-objective optimization problem. A case study was conducted in Hengshancounty of Shaanxi province in China to validate the model. The results show that the model features ahigher convergence rate and sampling accuracy than the traditional methods，and the combination ofpriori information and goal programming technique improves the representation of sampling solutionssignificantly. This model could be widely beneficial in soil sampling and monitoring.