Abstract: From late July to early August 2025, a regional major flood event occurred in the Haihe River Basin, with multiple flood processes observed in sub-basins including the Daqing River, Jiyun River, Chaobai River and Beiyun River. Leveraging the China Flash Flood Hydrological Model (CNFF), a retrospective analysis and calculation of floods were performed for key cross-sections within the basin, aiming to identify the sources of simulation errors. The results indicate that the CNFF model exhibits high precision and reliability at most monitoring stations across the basin, thereby proving its suitability for flood simulation in the Haihe River Basin. The primary contributors to model calculation errors include disparities in basin characteristics, inaccuracies in rainfall data input and variations in initial basin conditions. Based on the error analysis, model optimization should prioritize the following aspects: enhancing the spatial resolution of rainfall data to mitigate the smoothing effect; refining the unit division of large-scale basins; adjusting model parameters according to different underlying surface types and incorporating the influence of initial basin conditions into the simulation process.