Cloud microphysics play a key role in clouds and precipitation. The authors investigated the characteristics of clouds and precipitation, and synoptic conditions of a severe rainstorm in Beijing. The authors conducted sensitivity experiments for ten cloud microphysics schemes with the WRF (Weather Research and Forecasting) model in simulating the intensity, location, and duration of the rainstorm. Results showed that the severe rainstorm had an apparent short-duration and localized properties, and was a deep convective system formed by a merging of multicellular clusters, with favorable conditions for a large, mesoscale weather system and water vapor. Simulated intensity, location, and duration of the rainstorm in the model were very sensitive to the cloud microphysics scheme. The ETS (Equitable Threat Score) of the 18-h accumulated rainfall ≥50 mm and ≥100 mm for the different microphysical schemes indicated that only the Thompson scheme had both positive skill while other schemes had very low or negative skill, particularly for rainfall ≥100 mm. The complicated microphysics schemes had much better performance in the high horizontal resolution condition. Hourly precipitation intensity and duration simulated by the Thompson scheme were closest to observations, followed by the Lin and WSM6 schemes. Maximum area-accumulated precipitation was well-simulated by the Thompson and Morrison schemes. The single moment schemes of Lin and WSM6 better simulated the location of the precipitation, and the simulated precipitation amounts were much smaller. WDM6 scheme could well-simulate the area-mean precipitation, but had worse performance in the simulation of extreme precipitation. The differences were due to the different treatments of snow and graupel processes. Since size distribution, density, and terminal velocity of hydrometeors are usually differently parameterized in different microphysics schemes, this would induce different collision and formation processes of cloud particles. Most schemes produced higher graupel content and less snow content except for the Thompson scheme. Moreover, these differences also induced different feedback of cloud dynamics, primarily for the dragging process of precipitation particles on updrafts rather than the change of temperature profile due to release of latent heat in the phase change process of hydrometeors. Therefore, microphysics schemes including an appropriate parameterization of size distribution, shape, density, and terminal velocity of hydrometers would be beneficial to the improvement of heavy rainfall simulation in a cloud microphysics scheme.