Abstract: Purpose: A fast iterative image reconstruction method―OSEM (Ordered Subsets Expectation Maximization) is applied in the reconstruction of fan-beam projections of a Phantom, and the effective range of OSEM parameters (such as iterative number, subset order) is verified, and the OSEM with various subset orders are studied by comparing their reconstructive image quality and convergent speed. Method: The probability matrix of fan-beam geometry is calculated in advance and maximum interval between subset is assumed. Result: In the reconstruction of a simulated Phantom data, OSEM gives better images with higher spacial resolution and lower noise, as compared with traditional FBP method especially under noisy acquisition situation in fan-beam scanning, and in a certain extension the image quality is proportional to the product of the iterative number and the subset order, while computing time is only related to the iterative number. Conclusion: Good images with smaller iterative number and thus shorter time can be obtained by properly choosing the subset order in fan-beam OSEM reconstruction.