Abstract: A new method based on orthogonal polynomial expansion is proposed to reconstruct 3D CT projection in this paper. Firstly, a set of orthogonal bases in orthogonal space are used to expand the density function defined in cylindrical domain. The relation between density function and projection data is derived. And then, Gaussian quadrature rule is utilized to integral the above partial sum, which attains the reconstruction algorithm for 3D-projection data. Furthermore, the fast Fourier transform is introduced to improve the efficiency and feasibility of the proposed algorithm. Experiment results show that the algorithm proposed in this work can effectively handle the reconstruction task of 3D- projection with higher efficiency.