The accuracy and error characteristics of two dual-Doppler radar wind retrieval methods were evaluated by using Observation System Simulation Experiment (OSSE). A severe convective storm was simulated by the Weather Research and Forecasting model (WRF), and the model wind field was used as the true value to quantitatively calculate the errors of the retrieved wind field. Volumetric scans of reflectivity and radial velocity similar to operational radars were generated from the WRF output by placing virtual radars at different locations within the WRF domain. Three-dimensional wind fields were then retrieved from the simulated radar scans by the iterative and variational methods. The speed, divergence, vorticity, and vertical velocity fields in the retrieved wind were compared with the true wind at different periods of the lifecycle of the simulated convective storm. Results show that both methods were able to recover the basic characteristics of the wind fields. The variational method was able to better recover the vertical velocity, divergence, and vorticity fields of the convective storm, while those retrieved from the iterative method were either absent or too weak.