Abstract: Experimental studies of surface wind forecasts were carried out by employing the WRF model at two different resolutions of 10 km and 3.33 km on the background fi eld of global circulation forecasts from ECΜWF for 4 typical months in Liaoning-Jilin provinces. The conventional statistical verification, key wind speed segment verification, and different decomposition of forecast error were analyzed based on observation data at 47 wind towers in the two provinces during the same period. The verification results of effects of different horizontal resolutions on the forecast performance of wind speed in boundary layer were discussed in Northeast China region. The main conclusions are gotten as followings: 1) the effect of model horizontal resolution was little on the wind speed frequency, but the forecast performance was better in Liaoning province than that in Jilin; 2) The verification results of full wind speed segment show that the prediction performance at 70m height was better than that at 10m height; 3) the effect of model resolution on the RMSE after decomposition was little; but that for the phase difference was increased slightly; the system difference and amplitude difference were decreased slightly with higher resolutions; 4)The effect of different model resolutions on the forecast performances of the wind speed were little under different conditions of wind turbine generator. The forecast performances of wind speed under reactive condition were higher with a higher model resolution in winter and spring, while the forecast performance of wind speed under full load condition were higher with higher resolution in summer and autumn; 5) When the model resolution is higher, TS score was lower in the wind speed forecasts under power generation condition, the RMSE was higher and the relative error was higher in all seasons. In Northeast China, the forecast performance of wind speed was not better with higher model resolution.