The performance of the weather research and forecasting (WRF) model relies on the accuracy of the initial field provided by data assimilation. The initial field usually contains large uncertainties, especially for regions where observations are sparse or lacking. Assimilating additional observations is an efficient way to reduce this uncertainty. The moderate resolution imaging spectroradiometer (MODIS) is one of the most critical data sources of various indirect assimilating applications due to its wide swath and remarkable quality. Therefore, assimilating MODIS data into WRF data assimilation (WRFDA) system is meaningful to improve the accuracy of weather forecasts. This study developed a module to directly assimilate the MODIS radiances into the WRFDA based on the 3-D variational data assimilation method and community radiative transfer model. An assimilation experiment was carried out on August 2014, from which the background field has been relatively improved. Specifically, the improvement of the temperature, humidity, and wind speed at the near surface layer is about 0.2°C, 1.2%, and 0.2 m/s, respectively. Additional capabilities and increased potential of MODIS data assimilation based on WRFDA need to be further investigated and tested under various conditions and applications.