Source-sink landscape pattern is an important factor affecting the stormwater runoff process and flood risk in the watershed. Exploring a flood risk assessment method based on the source-sink landscape pattern is of great scientific significance for watershed flood risk prevention and control. Taking Dongting Lake Basin as an example, this study innovatively revised and applied the source-sink landscape model to assess the flood risk of the watershed every year, revealed the spatial and temporal evolution patterns of flood risk and causes. It was followed by using the SCS-CN model and the local isovolumetric method to simulate and analyze the characteristics of the stormwater runoff process every year. Then the Spearman's correlation coefficient was used to explore the correlation between the stormwater runoff process and the Location-Weighted Landscape Index (LWLI), in order to verify the rationality of using the new model in flood risk assessment. The results showed that (1) During a 40 years' timeframe, the expansion of construction land for source landscape was the most significant, while the sink landscape continued to decrease, with the main shift being from cultivated land to construction land. Source landscapes were spatially concentrated around Dongting Lake and the Xiangjiang River watershed. (2) The LWLI showed a rising trend during the 40-year period and a gradual increase of flood risk. This demonstrated a distribution pattern with high in the east and low in the west. Among them both the Xiangjiang River Basin and the Dongting Lake Zone showed a higher flood risk, followed by the Zi Basin. And the Li River and the Yuan River Basin had a lower flood risk. (3)The average runoff volume, the depth of inundation, and the percentage of each risk area during the 40-year period showed a slow fluctuating upward trend. The intensification of stormwater runoff process made the flood risk increase continuously. The results of the traditional model simulation coincided with the assessment results of the source-sink landscape model. (4) The LWLI and subindexes showed a significant positive correlation with each characteristic index of the stormwater runoff process, indicating that the pattern of the source-sink landscape in terms of elevation, slope and distance had an important influence on stormwater runoff process, and that it was feasible to assess the flood risk of the watershed by using the source-sink landscape model. This study could provide a new method of coupling pattern and process to quantify and analyze the flood risk in the watershed. It offered a new perspective and theoretical basis to rationally allocate the landscape pattern in the watershed and to protect the stormwater safety.