Abstract-A near-field three dimensional imaging algorithm for circular SAR is proposed in this paper. It adopts the theory of spherical wave decomposition to transform Green function to a superposition of plane wave components. Using this relation, the image-reconstruction can be implemented in frequency domain instead of in spatial domain, which simplifies the solving process of target reflectivity function, and allows for the target to be near to the radar. Through compensating phase factor and filtering at each elevation, we firstly get the ground CSAR signal of each elevation in frequency domain. Then, performing two dimensional inverse nonuniform fast Fourier transform and accumulating the results of all azimuth angles, the reconstructed two dimensional image corresponding to an elevation is achieved. Finally, using reconstructed image datum of all elevation, the three dimensional image of target is obtained. To demonstrate the imaging performance of our method, numerical simulations and experiments are conducted. By comparing the results with the focusing operator algorithm and the back-projection algorithm, it is found that the proposed algorithm is more efficient and can obtain a good imaging performance.