Abstract: According to the method of Extreme Forecast Index (EFI) developed by Lalaurette. Using the CMA T213 forecast precipitation data during 2007—2012, cumulative distribution functions (CDF) in Sichuan-Chongqing region was analyzed. An Extreme Precipitation Forecast Index (EPFI) model suitable for Sichuan-Chongqing region was established, then tested in three typical heavy rainfall processes. The results show that the EPFI can take advantage of the tail information of the CDF based on the EPS, providing scientific and rational forecast for extreme heavy precipitation. It can provide early warning information on extreme heavy precipitation 1 to 5 days in advance, and has the ability to correct or give guidance to the location of heavy rainfall. For different rainfall events, the ability of the EPFI are varied, but have a good performance on the falling area of extreme precipitation, and provide valuable information to the forecasters in the operational forecast.