Based on CWRF (Climate Extension of Weather Research and Forecast) model, 8 cloud microphysical parameterization schemes are compared in simulations of tropical cyclones (TCs) occurred over offshore areas of East Asia from 1986 to 2015 with respect to spatial distribution, frequency, and intensity. It is found that the TCs simulated by each cloud microphysical parameterization scheme in CWRF model have less frequency and weaker intensity compared with observations. The spatial distribution and frequency of the simulated TCs are sensitive to the selection of cloud microphysical parameterization schemes, whereas it has little impact on the TC intensity. The spatial distribution of the TCs is well simulated by the Morrison and Morrison-a schemes, while the annual variation tendency of TC frequency and intensity is poorly simulated by these two schemes in comparison with other schemes; by contrast, the GSFCGCE scheme performs better in the TS score and correlation coefficients between frequency and intensity. In all, the GSFCGCE scheme can better simulate TC activities than other microphysical schemes in CWRF model. Further analysis shows that compared with the Morrison and Thompson schemes, the Thompson-a and Morrison-a schemes, which take the impact of aerosol into consideration, can improve not only the simulation of frequency and spatial distribution, but also annual variation tendency of frequency and intensity.