A commercial computed radiography (CR) system was studied as an option for quantitative dosimetry quality assurance of external radiotherapy beams. Following the examination of influencing quantities, practical measurement procedures are discussed. Corrections were derived for image fading, an observed long-term response drift and the image length scale, which was found to be off by up to 2-3%. It is known that energy dependence is important for CR measurements. Therefore, signal-to-dose calibration curves and the energy dependence of the response were studied extensively using multiple photon and electron beam qualities. Doses which yield the same signal vary by up to tens of percent for different beam qualities. Results on the directional response of the plates are presented. It was found that rotations of up to 30° to 40° relative to perpendicular irradiation yield no significant change in response. Finally, the homogeneity of the response over the measurement region was studied for electrons and photons and a correction method is described. In summary, relative dose measurements with uncertainties of a few percent are feasible in regions of constant beam energy.