The WXGEN weather generator is a component of the soil and water assessment tool (SWAT), which is applied as a hydrological model at catchment scale and applied to forecast weather condition and fill the missing weather data. This paper analyzes and assesses the precision of the daily and monthly weather data based on the comparison between the data simulated by the WXGEN and the daily observed data according to 5 national meteorological stations in the upper Yangtze River. Results indicate that the WXGEN generated data are generally in consistence with those from the five stations. The errors of daily data simulated by WXGEN are significant, while the monthly data are more accurate than those of daily data. In conclusion, WXGEN has better simulation of the distribution of monthly weather data and hence it is more suitable for the simulation of monthly weather data in the Upper Yangtze River. Yet the errors of the simulated monthly weather data demonstrate significant seasonal changes. The simulated monthly precipitation and temperature are lower than the observed values, while the average monthly solar radiation is above the observed value.