[Objective] The study aims to investigate the spatial distribution of mountain torrent disasters in order to provide support for the prevention and management of mountain torrent disaster in overall Jiangxi Province and river basins.[Methods] Based on formation mechanism of mountain torrent disasters, nine explanatory variables were selected from triggering factors, disaster inducing environment and disaster-bearing body. Five response variables from the survey data of mountain torrent disaster were used as the indices to evaluate the hazard degree. The spatial patterns of mountain torrent disaster were estimated based on the geographically weighted regression (GWR) method. The similarities and differences of mountain torrent disaster in three different regions of Jiangxi Province were analyzed using GIS technology.[Results] There were similarities and differences among the models of different mountain torrent disaster indicators in the same area, the same mountain torrent disaster indicators in different regions and the spatial distribution of mountain torrent disaster indicators.[Conclusion] Not only the geographical differences but also the differences between different disaster degree indicators should be taken into account in building each model. GWR can effectively explain the local spatial variability and the differences of important explanatory variables.