Abstract: Risk assessment is necessary for mountain flood management. AHP-based linearly weighted addition approach was often used in existing assessment studies, but this approach has two obvious shortages. One is that some randomness is inevitable in the selection of hazard factors and the assignment of weight values, and another is that it is difficult for linearly weighted addition method to reveal the nonlinear relationship between hazard factors and mountain flood, leading bigger error. For this reason, a risk assessment method for mountain flood based on GIS and Logistic regression models is proposed, and an empirical study is carried out in Lushi County, Henan Province. The results show that:①The altitude, river network density and distance to main rivers are significant factors determining the spatial pattern of torrent hazard in Lushi County while the role of the slope was not important. Built-up land and cultivated land were more vulnerable; however, forest land, grassland, water, and vacant land were less vulnerable. ②River plains along Luo River, Qi River and Laoguan River, and nearby mountain areas were covered densely with high risk of mountain flood. Therefore, these areas should become hot spots for mountain flood management. In addition, other high elevation areas, far away from main rivers, were low risk for mountain flood. ③ GIS and logistic regression model based approach are effective for risk assessment of mountain flood hazard. It can accurately characterize the complex/nonlinear action of triggering factors to hazard system, and thus provide objective, scientific information to evaluate mountain flood risks.