[Objective] The spatial distribution prediction of soil total nitrogen and its influencing factors are essential for understanding nitrogen cycle and formulating soil nitrogen management measures. Combined with measured soil total nitrogen content data and multi-source environmental variables, the random forest model was established to explore the spatial distribution characteristics of soil total nitrogen content in a forest farm and the response to environmental factors. [Method] The Wangyedian Experimental Forest Farm was taken as the research area, and a quantitative relationship was established between soil total nitrogen content and environmental covariates of altitude, normalized vegetation index, slope aspect, mean annual temperature, mean annual precipitation and y coordinate at soil depths of 0-10, 10-30 and 30-50 cm using the random forest model. Soil total nitrogen content was predicted and mapped, and the controlling factors affecting its spatial variation were analyzed. [Result] The results showed that (1) the average soil total nitrogen contents at soil depths of 0-10, 10-30 and 30-50 cm were 3.20, 2.02 and 1.47 g kg-1, respectively. It decreased with the increase of soil depth. (2) Cross-validation results showed that the R2 of soil total nitrogen at the three soil depths were 0.54, 0.40 and 0.38, respectively, and the prediction accuracy of the model decreased with the increase of soil depth, indicating that the random forest model could explain 38%~54% of the spatial variation of soil total nitrogen. The effects of various environmental factors on the spatial distribution of soil total nitrogen decreased with the increase of soil depth. (3) Judged from the prognostic map, soil total nitrogen content in different soil depth showed a spatial pattern of low content in the west, north and central parts, and high content in the southwest, southeast and east parts. (4) Altitude had the greatest effect on soil total nitrogen content, and the rest was in the order of normalized vegetation index > mean annual precipitation > y-coordinate > aspect > mean annual temperature. [Conclusion] The random forest model can effectively predict soil total nitrogen content at different soil depths in this forest farm, providing a new idea for the spatial distribution prediction of other soil attributes. Meanwhile, this study provide a reference for forest soil nutrient management and sustainable use on forest farm scale.