Taking Typhoon In-Fa (2106) and Chanthu (2114) as examples, this study investigates whether the multi-model QPF (quantitative precipitation forecast) fusion method can add values to the predicted 6-h heavy precipitation (R≥25 mm) of typhoon rainstorm in Zhejiang. The model forecasts include precipitation and typhoon track from ECMWF (European Centre for Medium-Range Weather Forecasts) typhoon track ensemble, CMA-MESO 3 km, CMA-MESO 10 km, CMA-SH9, and ZJWARMS (Zhejiang WRF ADAS Real-time Modeling System). The analysis results are listed as follows. (1) For the two typhoon precipitation processes, all of the 4 regional models overestimate the rainstorm in Zhejiang. In contrast, the multi-model QPF fusion method can effectively improve the ETS (equitable threat score) and reduce the false alarm ratio of rainstorm forecast. (2) Compared with the best model for forecasting Typhoon In-Fa (2106), i.e., CMA-MESO 3 km, the multi-model QPF fusion method can increase the POD (probability of detection) of rainstorm and heavy rainstorm by 18.80% and 23.41%, and increase the ETS of rainstorm and heavy rainstorm by 24.37% and 25.76%, respectively. Similarly, compared with the best model for forecasting Typhoon Chanthu (2114), i.e., ZJWARMS, the new method also improves the ETS of rainstorm and heavy rainstorm by 23.08% and 3.23%, respectively. And both of the method’s ETS and the POD are higher than the operational objective forecast of Zhejiang in the two case studies. (3) Furthermore, when there are large differences in the typhoon track forecasts between several regional models, the multi-model QPF fusion method can significantly increase the accuracy of typhoon rainstorm forecast.