More than 300 diagnostic x-ray units are subjected to radiation protection surveys and calibrations annually in a medical physics practice involving more than 40 diagnostic facilities ranging from small private offices to several hospitals having 500-700 beds. In addition to traditional protection oriented analysis, performance characteristics are evaluated in order to improve quality assurance and reduce the needless patient exposure and cost associated with repeated examinations. A long term study, previously reported by the authors, includes development of ionization techniques for and results of analysis of linearity of exposure, mR/mas constancy, as functions of indicated time and current. Determination of accuracy of kVp was not done in the past because the cost-benefit ratio was considered excessive using conventional techniques such as voltage dividers in large scale measurements. The recent availability of the RMI Wisconsin X-ray Test Cassette based on the Ardran Crooks technique has made it possible to provide kVp accuracy tests as a part of the routine annual survey at minimal cost. During 1975-76, determinations of kVp accuracy have been made on more than 100 stationary radiographic x-ray units at 60,80, 100, and 120 kVp. These data have been analyzed in terms of the frequency distribution of actual/indicated kVp ratios. For 66% of all units the measured potentials were within + 5% of the indicated kVp. All focal spots measured were equal to or less than the specified size. We propose as practical acceptance test criteria that mR/mas variation on each unit for all currents and times (resulting in greater than 1 mas) be within ± 10%, that kVp be within ±5% of indicated kVp, and that focal spot size be within ±10% of that specified.