Abstract: Quality and time of the image reconstruction with algebraic reconstruction techniques(ART) algorithm is influenced by many factors. The influences of some factors including the prior knowledge, the initial value, different projection numbers under different noise levels and projection data with limited angles on quality of the image reconstruction and selection of relaxation parameter, were investigated with computer simulation experiments. The experiment leads up to the result that selecting the proper relaxation parameter and adding the priori knowledge can greatly improve the image quality. At the same time, the priori knowledge and different projection numbers under different noise levels also impact the selection of optimum relaxation parameter. Under the circumstance of projection data with the lack of a certain range of observation angles, the quality of image can be improved by evenly increasing the projection numbers.