To expand torrential rain, which is a meso- and microscale weather process, to a meso- and long-scale weather process, in this paper, we choose South China as a sample region and propose the conception of the “Cumulative Effect” of torrential rain (CETR) by using daily precipitation observational data from 740 stations. Through a statistical analysis of the observations, three indexes—continuous time (L _d), control area (A _r), and precipitation contribution rate (Q _s)—are used to define the CETR and indicate the torrential rain processes. The relationships between the CETR and simultaneous total precipitation over South China are analyzed in the pre-flooding and latter flooding seasons. This analysis shows that on both interannual and interdecadal scales, the three indexes are highly correlated with simultaneous total precipitation over South China in the pre-flooding season and latter flooding season. Moreover, an empirical orthogonal function (EOF) analysis is performed to classify the spatial distribution of the CETR. In both the pre-flooding season and the latter flooding season, the four major spatial models of torrential rain are similar to those of total precipitation over South China. With regard to the amount of precipitation caused by the CETR, the latter flooding season is affected more significantly than the pre-flooding season. Regarding the geographical distribution of precipitation, the opposite result occurs. In conclusion, in both the pre-flooding season and the latter flooding season, the CETR influences and even determines the amount and distribution of precipitation over South China.