With the development of flat panel detector technology, cone-beam computerized tomography (CBCT) has become an important image technique, and been widely applied in clinical practices. Specially, C-arm mounted CBCT has shown extra suitability in image guided interventional surgeries. However, how to acquire real-time, high resolution, and high quality 3D images remains challenging. In this paper, a GPU based acceleration method for fast C-arm CBCT 3D image reconstruction was proposed. With the acceleration both in algorithm and system design, it is shown that the system performance is enhanced with the proposed system design by 26% and reconstruction delay is accelerated by 2.1 times when 90 frames of projections were used, and the performance increases to 39% and 3.3 times with 120 frames.