Based on hourly observations of wind speed at Nanjing station and three automatic weather stations in the surrounding rural area in 2005, the probability distribution function of wind speed is fitted in this study. Analysis indicates that the 3-parameters Weibull distribution well fits the urban-rural surface wind speed distribution, while significant differences can be found in curves of wind speed probability density function (PDF) between urban and rural areas. The curve of wind speed PDF is more precipitous in urban than in rural area, suggesting that the wind speed distribution is more concentrated in urban area. PDF values are significantly higher in urban area than in the surrounding rural area for wind speeds within the range of 0.75-3.75 m/s, but lower for wind speeds beyond this range. Frictional effect generated by urban underlying surface can weaken the wind speed while the urban thermal effect can intensify the wind speed. Analysis of time series of urban-rural wind speed difference indicates that most of the time urban wind speed is smaller than rural wind speed, but larger while the wind speed is lower than 1.9 m/s. The impact of urban surface frictional effect is much greater than that of its thermal effects in general, and urbanization leads to a decrease of 0.43 m/s in annual mean wind speed.