Abstract: The precipitation Extreme Forecast Index (EFI) derived from ECMWF global ensemble prediction system (EPS) from June 2016 to May 2017 is applied to evaluate the statistical relationship between the EFI and the severe precipitation as well as the climatic precipitation percentile in Zhejiang. Results show that there is an obvious positive correlation between the EFI and the observed precipitation in Zhejiang. With the increase of the EFI threshold, the frequency of rainstorm increases at first and then decreases, and the probability of rainstorm increases with the increase of EFI threshold. With the increase of the EFI, the frequency of rainstorm increases at first and then decreases, while the probability of rainstorm increases with the increase of the EFI. Considering the TS and BS scores comprehensively, the threshold of the EFI decreases with the increase of forecast time length and increase with the increase of precipitation. There is a significant positive correlation between the EFI value and the climatic percentile of precipitation. The greater the EFI is, the higher the percentile is. Thus, the precipitation amount can be estimated according to the corresponding climatic percentile of the observed precipitation.