Abstract: In order to use quality-controled wind profiler network data (hereinafter referred to as observation) to evaluate the prediction ability of the horizontal wind in Wuhan Mesoscale Model (WHMM), comparison results of one month (May 2013) in the whole domain, and at different heights, speeds and single stations are analyzed, respectively, by calculating the correlation coefficient, bias, and root mean square error. Results show that (1) WHMM has a good performance for the horizontal wind forecasting. The correlation coefficients of 12 h and 24 h forecasts are larger than 0.6 and correspond to α=0.01 significant level. The correlation coefficients of 12 h forecast are greater than that of 24 h, and there are negative biases in the wind speed forecast, which is smaller than the observations.With the increase of the forecast range, the forecasting errors of the wind increase. The prediction of the U-component (u) and the wind speed (wspd) are better than that of V-component (v). (2) In the vertical distribution, the RMSEs of wspd, u and v of WHMM forecast increase first and then decrease with height. The RMSEs at the height of 1~2 km are largest. The RMSEs at the height of 4~5 km are smaller than that at other heights. (3) In groups of ≤ 5 m·s^-1, 5-10 m·s^-1, 10-15 m·s^-1, 15-20 m·s^-1 and 20-30 m·s^-1, the correlation coefficients between 12 h predictions and observations are larger than those at 24 h, and they all correspond to α=0.01 significant level test. The RMSEs of wspd, u and v of WHMM forecasts increase with the increasing wind speed under different wind speed groups. (4) From wind profiles at a single station, the model can predict the trend of wind field over the station with height. The forecasting skill at Guangzhou station is better than that at Wuhu and Zigui stations.