Abstract: Based on NCEP reanalysis data and the observation data of thunderstorms at national and regional stations from May to August between 2007 and 2015, using the method combining weights and probability statistics, the physical quantities that are closely integrated with the thunderstorm were determined, and the probability within different thresholds was calculated. Finally, thunderstorm potential prediction equation was established, and the forecast results were tested. The results showed the following. 1) Physical quantities selected are different between day and night. Meanwhile, whether it is during the day or at night, thresholds of physical quantities are also different between mountain areas and other areas. 2) The probability of each selected physical quantity appeared to be basically consistent with environmental conditions of thunderstorm, which was determined by the number of samples of physical quantities and thunderstorms within the threshold. 3) The method showed a good forecast of thunderstorm in from 08:00 to 20:00 BT and from 20:00 to 08:00 BT, especially in July 2016, during which the hit rate of more than '60%' was close to 93%. It can well predict the potential and possible areas of thunderstorm. At the same time, its vacancy rate was also high. The critical success index of '80%' was even higher.