To improve the image reconstruction resolution of ultrasonic tomographic flow field imaging, an iterative Filtered Back Projection (FBP) algorithm was proposed. On the basis of the Simultaneous Iterative Reconstruction Technique (SIRT), the FBP algorithm was integrated into an iterative reconstruction process. With the FBP algorithm, an error image was reconstructed by the residual error of the projection data, then the flow field image was revised and the iterative reconstruction of the image was implemented. By optimizing the iteration step size, the residual error of the projection data for each iteration was minimized to improve the speed of convergence. Finally, according to the flow field characteristics in continuity and compact support, the projection data was interpolated and the flow field image was assigned a value of zero outside the circular domain. Simulation results indicate that the iterative FBP algorithm decreases the image error by 26% and the volume flow rate error decreases from an average of -1.77% to ±0.25% as compared with the FBP algorithm. In addition, the computational time of the iterative FBP algorithm is just 0.63 s, only 0 89% that of SIRT. In actual experiments, the iterative FBP algorithm can reconstruct the flow fields in a straight pipe and downstream of a 90° single bend with high reliability. It concludes that the iterative FBP algorithm implements the reconstruction of flow fields in real-time and higher accurate because of its good reconstruction resolution and high computational efficiency.