Standard precipitation verification method, extreme dependency index and method for object-based diagnostic evaluation (MODE) are applied in this paper to make evaluation of SMS-WARMS V2.0 precipitation in Southwest China from December 2015 to November 2016. The results show that (1) the treat scores of four seasons in Southwest are high, especially in spring and summer, and the performance is relatively stable within 48 h. (2) The results of frequency bias and true skill score show each magnitude precipitation in spring and summer forecasted by SMS-WARMS V2.0 in Southwest China is more than observation, but the model forecast to torrential rain in autumn and heavy rain in winter is relatively less. Overall, the forecast skill in Southwest China exhibits the features that the POD is higher than FAR. (3) The treat scores of light rain, torrential rain and downpour forecasted by SMS-WARMS V2.0 are higher than ECMWF model. (4) Model has a good performance in forecasting extreme precipitation in Southwest through extreme dependency index, especially better for central and northeastern Sichuan and southwestern Guizhou. (5) Precipitation area forecast of a rainstorm process caused by a southwest vortex in August 2015 performed well in space, but the intensity was stronger than observation. (6) Statistic results made by MODE show that the rainstorm object centroid has a small bias between forecast and observation, but the central intensity of precipitation is stronger than observation.