This paper reviews the progress in observation, numerical simulation, dynamic mechanism, diagnostic analysis and prediction method for three major rain belts in recent years from the perspectives of heavy rainfall in warm South China, rainstorms caused by the Meiyu front in the Yangtze-Huaihe River basin, torrential rain in North and Northeast China and forecasting methods for rainstorm. This paper summarizes some new understandings of intense rainfall in recent researches. First of all, it is found that rainstorms in the warm sector of South China mainly occur during the process when low-level southerly winds move northward, which is the major mechanism for rainstorm formation in South China. During this process, the potential instability caused by coastline and orographic lifting can lead to strong ascending motions. Besides, the northward movement of low-level southerly winds may converge in front of the jet stream, leading to streamline forks that are favorable for the development of cyclonic vortices or the concentration of positive vorticity and promote vertical motions, which often cause torrential rainfall. The mechanism of heavy rainfall over the Yangtze-Huaihe River basin is mainly associated with symmetric instability and variation of vorticity field as well as the β mesoscale convection line. In North and Northeast China, local concentration of humidity is more obvious due to the non-uniform saturation during rainfall process. The instability and downward momentum transport caused by dry and cold air in the middle and upper levels play a significant role in rainstorm generation, while the enhancement of local upper-level divergence caused by the reinforce of mesoscale upper-level jet is also important. Recently, new sounding data have been used in rainstorm research and prediction. The GRAPES-MESO and GRAPES-GFS systems with independent intellectual property also have been put into operation, and make impressive progress in numerical ensemble prediction. In addition, the dynamic-factor approach to predict rainstorm has been popularized and applied in many provincial and municipal meteorological bureaus. Although great progresses have been achieved in the research and forecast of rainstorm mechanisms, there still exist a lot of problems. For instance, it is crucial to promote theoretical research of meso-scale rainstorm systems based on observations. Further improvements of dynamic framework and better descriptions of physical processes in numerical models and new developments of theory and technology of data assimilation are also important. Moreover, how to apply artificial intelligence technology to atmospheric research and business, how to refine the three-dimensional structure of rainstorms, how to further explore mechanisms for the generation and development of rainstorm, and how to advance the prediction theory for heavy rainfall mesoscale system all need to be considered in the future study.