PURPOSE: (1) Do radiographic protocols for knee axial images vary across different clinic locations? (2) Does variability between radiographic positioning protocols (knee flexion angle, beam positioning) yield different radiographic measurements that could influence clinical assumptions? METHODS: Radiographic positioning, including beam angle and knee flexion, was measured using a photograph taken of a phantom leg and (human) model set-up. An axial radiograph of a phantom leg was taken at seven different outpatients musculoskeletal clinics to assess patella position (lateral tilt/translation). The phantom leg patella was placed in an abnormal position for the radiographic image across all sites. RESULTS: Knee flexion<30° was less accurate in obtaining requested knee flexion than higher knee flexion angles (phantom limb error averaged 6°, range 4°-7° human model error averaged 15°, range 11°-25°). Of the five sites utilizing a 'Merchants' axial radiographs, the congruence angle varied from +12° to +29°, being normal at one site (+12°) and abnormal (>+16°) at three sites, and negative at one site (-15°). CONCLUSIONS: Radiographic protocols for knee axial imaging are inconsistent in radiographic positioning and imaging. This variability can lead to disparity in radiographic image results used for clinical decision-making. Standardization of radiographic positioning would yield more consistent imaging of patella position in early flexion, increasing clinical utility and accuracy.