One method of detecting change in an anatomical structure is by comparing elements of a longitudinal series of planar transmission images (i.e. radiographs). Unfortunately, changes in image formation parameters, such as patient placement, prevent successful image registration from occurring. In this paper, an analytic method is described which allows one to quantify the ability to register two planar transmission images. The concepts described in this paper are not dependent on a specific image registration method. In fact, these results hold independent of the registration algorithm used since they are based solely on phenomena occurring during the image formation process.