Abstract: The Jinyun Mountain forest fire event in August 21 to 26, 2022 has attracted extensive attention in China. The forest fire hazard has a long-term impact on the nature soil of forest plant and soil. In addition to the mechanism and prevention measures of forest fire, the safety prediction for post-hazards (such as landslide) can also be one of the most urgent tasks at present. But now, the main challenge is the long-term estimation of dynamical root reinforcement for the lack of monitoring data on plant root death and regrowth post fire. This study aims to clarify the mechanism of forest slope failure, and then to predict the landslide risk post fire. The field and lab tests were performed on the physical properties of root soil in the burned woodland in the Hutou Village near Jinyun Mountain. The SIDLE curve model was selected to predict the dynamics of root reinforcing strength after forest timber harvesting. Finally, the simulation was conducted for the post-40 years slope stability using COMSOL. Specifically, a double-layer slope model was assembled, according to the surveyed slope size and soil property. A systematic analysis was made on the root strength time inflection points of common trees and shrubs in published literature and according to which the SIDLE parameters of shrub in this work were inversed and dopted for simulation work of slope deformation and stability in the future. Both the general and extreme rainfall conditions with an intensity of 80 mm/d were considered in the simulation. The results show that: 1) The common type of forest margin slope in Jinyun Mountain was characterized by the double-layer soil slope structure consisting of sandy loam and deeper sandy mudstone. There was a significant difference in the magnitude of average conductivity and shear strength parameters between the upper sandy loam and lower layer sandy mudstone. The conductivity of the surface layer was 5.2×10⁻⁶ m/s on average, 3.47 times that of the lower layer. The cohesion and internal friction angle of the surface layer were 5.5 kPa and 18°, respectively, which were 3.06 and 0.69 times that of the sublayer, respectively. 2) In the general situation without rainfall infiltration, the slope was in a stable state with a potential sliding surface located in the deep layer (F_OS>2), while under extreme rainfall conditions, a waterlogged zone was prone to appear on the upper and lower interface with the positive pressure of 10 kPa. The factor of the safety of the bar slope (F_OS=1.13) was lower than that of the vegetated case (F_OS=1.32), indicating the key role of plant roots for soil strength compensation and slope stability. 3) Within a few years after the fire, there was less impact of the fire event on the stability in general situation, because the critical sliding surface was located in deep soil that was far below the root zone. By contrast, the potential sliding surface was quite shallow under extreme rainfall conditions. As such, the F_OS versus annual time presented a ‘rapid decreasing and then slowly increasing’ shape curve, indicating a landslide-prone period last for several years. Consequently, a dramatic variation was found in the strength and permeability of the upper and lower layers of soil under the common type of sandy loam-mudstone slope in Jinyun Mountain. A waterlogged zone was prone to appear on the soil interface under extreme rainfall conditions, leading to the reduction of local effective stress. This was the main reason for the slope failure. In current time, the F_OS of bar slope and vegetated slope are 1.13 and 1.41 respectively under the extreme rainfall condition, owing to root reinforcement of 19.8% contribution. Taken the root decay and recovery process into account, it is conservatively estimated that the landslide-prone period occurres in 6-9 years after the fire event with a F_OS between 1.22 and 1.32, while the slope recovered to the initial stability level of at least 18-24 years. This preliminary prediction can also be required long-term observation and verification in the future.