Abstract: To meet the requirement of developing a new method for evaluating the forecast skill of heavy rainfall, the main factors affecting forecaster's confidence in heavy rainstorm forecasting, that is, the forecasting ability of statistical characteristics of the heavy rainstorm climate, the characteristics of movement scale of systems influencing heavy rainstorms and the numerical prediction model, are considered in the present study to develop a new mathematical model of Synthetic Predictability Index of Heavy Rain (SPI). The SPI is composed of three components:rainstorm climatic frequency, rainstorm area ratio and numerical model rainstorm forecasting success index (Threat Score, TS). It is established based on analysis of 5 km×5 km resolution multi-source precipitation fusion grid analysis data, precipitation observation data at weather stations, precipitation forecast data produced by operational model on regional scale and the statistical method of extended space rainstorm samples of the National Meteorological Information Center. The three components of SPI and their spatial-temporal variations during April-October from 2008 to 2016 are calculated. The results show that heavy rainfall changes with season and its spatial distribution is not uniform. From April to May, the more predictable areas are mainly distributed in southern China; from June to July, the more predictable areas are mostly located in the Changjiang-Huaihe river basins; from mid-July to August, they are largely found in North and Northeast China. In September, following the southward retreat of the subtropical high pressure, the large value center of SPI moves southward correspondingly. In addition, the partial correlation coefficients between the rainstorm predicta-bility index and the three components shows that the partial correlation coefficient between the SPI and the storm area ratio is the highest with the value higher than 0.9. The comprehensive index of rainstorm predictability in China has laid a footstone for the development of verification scores of rainstorm forecasting based on predictability.