A description is given of the different techniques for height extrapolation of wind speed. A study was conducted on the sensitivity of the similarity model and modified power law model to different input parameters. The models are most sensitive to wind speed at anemometer level and to net radiation (atmospheric stability) and are less sensitive to surface roughness length. Data were also studied from sites selected as candidates for installation of large DOE wind turbines. These candidate site data, given in wind speed frequency form at two or three altitudes, were used to test and compare the modified power law model with the empirical models.