This paper puts forward a method to improve the traditional model of regional flood risk assessment with real-time remote sensing monitoring data and historical flood disaster data. It takes the Songhuajiang River Basin as the study area to build the improved regional flood risk assessment model. According to the characteristics of the study area, factors such as rainfall, topography, water density, population density and GDP are selected to build the regional flood risk assessment model based on GIS and AHP, by which the initial Flood Risk Index of Songhuajiang River can be calculated. The historical flood level distribution data that reflect the spatial distribution and frequency of flood events in the study area can be used to improve the hazard index. And with the real-time remote sensing data, the real-time flood extent data can be extracted to improve final risk assessment which reflects the regional flood risk better.