Abstract: Benggang (collapsing hill) is one of the most severe forms of soil erosion in the red soil region of southern China. This typical erosion is also characterized by sudden damage. Therefore, it is extremely necessary to assess the risk of Benggang erosion for subsequent comprehensive rectification. In this study, the Driving Force-Pressure-State-Impact-Response (DPSIR) model was applied to assess the risk of Benggang erosion, together with the natural disaster risk theory. A comprehensive system of risk assessment was constructed from the perspectives of susceptibility, vulnerability, exposure, disaster prevention, and mitigation capabilities. The case area selected was Guanqiao Town located in the southeast part of Anxi Country, Quanzhou City in Fujian Province, China. First, the spatial distribution of Benggang erosion risk was obtained using the ArcGIS spatial analysis for the years 2010, 2015, and 2020. The transition probability matrix of Benggang erosion risk class in the two periods of 2010-2015 and 2015-2020 was then calculated to characterize the temporal and spatial evolution of risk changes. Last, the relationship between the comprehensive evaluation and simple susceptibility was further analyzed to compare the spatial distribution. The results indicated that: 1) The DPSIR model was suitable for the comprehensive risk assessment of Benggang erosion in the study area, indicating high accuracy (the AUC of the three years were all greater than 0.59). Among them, gully density performed the most on the Benggang erosion risk (weight of 0.217), followed by soil erosion intensity and population density (weights of 0.182 and 0.174 respectively). 2) The risk of Benggang erosion was mainly low-risk grade areas ( accounting for 57.49%-62.90% of the total area ), with the distribution pattern of high-risk areas in the central and low-risk areas in the surrounding areas. The changes in the risk grades were distributed mainly in the central-western and eastern regions. The medium-, high-, and extremely high-risk grades were transitioned to lower-risk grades. However, the spatial change rate slowed down from 2015 to 2020. 3) Comparative analysis showed that there were broadly consistent trends of spatial distribution between the evaluation and simple susceptibility. Nevertheless, larger spatial variability was found in the comprehensive risk assessment. Therefore, the comprehensive Benggang erosion risk can significantly contribute to a better understanding of the influence of different indicators in the erosion risk system. The findings also provide an overall strategy and technique guideline to prevent, control, and manage erosion risk in Benggang regions.