There are kinds of physical processes in the mesoscale model, in which, the microphysical process is of vital importance. It makes direct effect on the dynamical process of the model and the precipitation forecast as well. Different physical schemes have different physical basis, in the simulation of torrential rain, they will produce distinct results. It needs detailed comparison to justify which scheme has better performance in the heavy rain simulation. A heavy rainfall in North China during 18-19 August 2010 is simulated using the weather research and forecast modeling system (WRFV3.2.1 version). The study is carried out with seven microphysical schemes with the 36-km, 12-km, and 4-km resolutions, respectively. From the rainfall area and intensity of precipitation,forecast performances (model performance) are compared. The simulation results show that different microphysical schemes can result in different degrees of simulation effect including the scope and intensity of the rain. Reasonable selection of microphysical schemes can also improve the fine-mesh (4 km) nested simulation accordingly, thereby improving the resolution of the rainstorm simulation, which provides reference in the analysis of small- and medium-scale rainstorm. The project with the grid size of 36 km indicates that: by comparison, the simulated rainfall location and intensity with the Lin scheme are closer to observed ones. For 12-km resolution, Thompson scheme performs better than other schemes in simulating both the heavy rainfall center and strength. When the grid resolution reduced to 4 km, WSM6 scheme performs best. Combined with the diagnosis of the basic physical quantities, such as vertical speed, vorticity, divergence, and the rain mixing ratio, the role that microphysical schemes play in the rainfall forecast can be better understood. The conclusion is of certain reference value in the rainfall forecast and the mesoscale model microphysical processes in operation and research.