Abstract: Reasonable arrangement of soil sampling points is important to fully grasp the spatial variation characteristics and trends of heavy metals. At present, meshed uniform sampling is still the main method in soil heavy metal researches. However, these methods fail to fully consider the cost and representativeness of sampling and cannot meet the requirements of high-precision spatial analysis methods. In this paper, the heavy metal sampling layout was carried out based on human influencing factors. By screening the human influencing factors affecting soil heavy metals in the region, the spatial generalization of influencing factors was carried out by using the nuclear density method. With the optimal semi-variational variance as the objective function, the sampling scheme of sampling points was analyzed through the range of variation. The north plain area of Longkou was taken as a typical research area, and two important factors, namely industrial and mining enterprise density (DI) and road traffic density (DR), were selected as the prior knowledge to seek the optimal sampling point collection scheme for soil Hg element. Finally, the effectiveness of the proposed layout scheme was verified by sampling based on different intervals. The results showed that (1) DI and Hg showed similar spatial distribution trends in the southern and northeastern regions of the study area, while DR and Hg showed relatively similar spatial distribution trends in the southern and southeastern regions of the study area. (2) The effective range length of DI and DR was 5819 m and 6079 m, respectively, which was also similar to the range lengths of the verification data Hg (6000 m). (3) In order to verify the variation range of human factor index can be used as the reference value of soil Hg sampling distance, the actual Hg sampling scheme with different sampling spacing (3000 m, 4000 m, 5000 m, 6000 m and 7000 m) was set, and the Kriging interpolation accuracy (MSE, RMSDE) was used to obtain the estimation error of heavy metals under different sampling schemes. By comparing the errors of different sampling intervals, it is found that when the sampling interval is less than 6000 m, the errors of different sampling intervals are almost the same, but when the sampling interval is 7000 m, the errors increase significantly. Therefore, when 6000 m is used as the sampling interval, the cost and representativity of sampling can be fully considered and the requirements of high-precision spatial analysis method can be met. Moreover, this value is the same as the range (6000 m) calculated by DI and DR, which further proves that the range calculated by human factor indicators DI and DR has reference value for the layout of Hg sample collection scheme.