Abstract: High resolution single photon emission computed tomography (SPECT) is a very important technology in clinical and small animal imaging. It has been proved that by modeling the system matrix of SPECT and using the iterative reconstruction algorithm, the spatial resolution can be improved efficiently. In this research, a method which combines Monte Carlo simulation and analytical model to calculate accurate system matrix was studied based on the small animal SPECT system. After the simulated point source projection data were acquired, the feature parameters of point spread function (PSF) were extracted by Gaussian fitting. Using the analytical model and the parameters make calculating the accurate PSF of any position in this system possible. In the reconstruction results of using two different system matrixes, one only used the analytical model and the other used the new method. The results validate that the method proposed in this paper improves the spatial resolution of image. Furthermore, the influence between different point source arrangements was studied in order to acquire a better arrangement. With respect to the method based on the experiment, this method can accurately realize the modeling of system matrix and improve the spatial resolution of image in a more appropriate way.