Directly assimilating QuikSCAT surface winds has resulted in only a marginal impact in hurricane initialization and prediction. In order to maximize the impact of QuikSCAT observations on hurricane prediction, a radial profile of sea level pressures (SLP) is first derived from QuikSCAT surface winds, and then a four-dimensional variational data assimilation procedure is used to assimilate the QuikSCAT-derived SLP. Specifically, radial profiles of QuikSCAT surface winds are first calculated in the four quadrants of the hurricane (north-east, south-east, south-west, and north-west). Second, radial profiles of wind are corrected by Holland's formula. Third, radial profiles of SLP are derived using the gradient wind relation. Finally, the SLP is assimilated in a half hour window under the forecast model constraint, which creates a realistic initial vortex in all model variables. The fifth generation Pennsylvania State University and National Center for Atmospheric Research Mesoscale Model (MM5) and its adjoint modeling system are used for this study. The impact of QuikSCAT winds on hurricane initialization and forecasts are examined for three hurricanes: Alberto (2000), Gordon (2000) and Erin (2001). Numerical results will be presented at the conference.