Based on traditional Time Variant Gain Model (TVGM), this paper developed a multi-source TVGM (MTVGM) by considering the additional baseflow process. MTVGM is composed of surface flow submodel and baseflow submodel, which independently calculate the runoff generation and flow concentration respectively. The model was then applied to 13 subbasins above Bengbu water sluice in the Huaihe River Basin. The results show that MTVGM can not only keep the advantage of the TVGM which uses the simple structure to describe the complexity in nonlinear hydrological relationship, but also greatly improve the forecast accuracy. Among the parameter calibration and model verification period, the forecast accuracy of 9 and 7 subbasins respectively meet the B -grade for traditional TVGM, while the subbasin numbers increase to 12 and 8 respectively for MTVGM. The performance of MTVGM was especially improved in large basins and large flood years. The water balance efficiency improves from 0.88 of TVGM to 0.95 of MTVGM, and the average increase is about 7%. The relative flood peak error is also promoted from 50% in TVGM to 92.3% in MTVGM, with an improvement of 42.3% in total. Therefore, MTVGM can be better applied in flood forecasting in Huaihe River Basin.