Abstract: A method for registering acquired images from a prototype flat panel mobile C-arm, capable of kilovoltage (kV) cone-beam computed tomography (CT), to a linear accelerator (LINAC) isocenter is presented. After analysis of algorithms capable of calculating transformations, one using singular value decomposition (SVD) was selected as the best choice in terms of speed, accuracy, and implementation. The algorithm requires accurately locating a set of points in two coordinate systems. Two distinct methods were developed to accomplish this. The 'tracking method' (TM) involved locating reflective markers using both the image capabilities of the C-arm and a commercial optical tracking system, the 'imaging method' (IM) required location of radio-opaque fiducial markers. An accuracy evaluation of the methods was accomplished by locating markers on a test phantom. The positions of the markers in world/room coordinates according to the optical tracking system were compared to their location in world/room coordinates based on a reconstructed image and a transformation generated by one of the two methods. The TM showed a consistent accuracy comparable to systems routinely used for image-guided radiation therapy (IGRT). The IM showed a lower accuracy with larger variability, believed to be caused from the digital readout on the treatment table, relied upon for location of fiducial markers. The merit of these methods was evaluated by applying them to patient setup applications. Real-time shifts were calculated with both methods and applied to an anthropomorphic phantom; aligning an embedded BB at the LINAC isocenter. Using the TM method produced more accurate results; equivalent to current IGRT systems. The procedure times of both methods were comparable to other IGRT systems, but currently the IM takes twice as long as the TM. Recognizing the advantage of comparing patient image sets to planning CTs resulted in the creation of a technique capable of transforming the entire image set into world/room coordinates. Evaluation of the technique demonstrated minimal added uncertainty. The technique was used to develop a complete procedure to compare patient planning CTs to C-arm images for setup evaluation. Time restrictions make the procedure impractical for real-time patient setup; however, off-line studies might provide useful.