X-ray technology has been widely used in a number of industrial applications for monitoring and inspecting inner defects which can hardly be found by normal vision systems as a ball grid array (BGA) or a flip chip array (FCA). Digital tomosynthesis (DT) is one of the most useful X-ray cross-sectional imaging methods for PCB inspection, and it usually uses an X-ray image intensifier. However, the image intensifier distorts X-ray images severely both of shape and intensity. This distortion breaks the correspondences between those images and prevents us from acquiring accurate cross-section images. Therefore, image distortion compensation is one of the most important issues in realizing a DT system. In this paper, an image distortion compensation method for an X-ray DT system is presented. It is to use a general distortion polynomial model on two dimensional plane that can cope with arbitrary, complex and various forms of distortion. Experimental results show a great improvement in compensation speed and accuracy.