Background: Humeral retroversion (RV) is important to the study of shoulder function and reconstruction. This study tests the hypothesis that clinically obtained computer tomography (CT) measurements for humeral RV (off-axis measurements) differ from those obtained after reformatting the image slice orientation so that the humeral shaft is perpendicular to the gantry (coaxial measurements) and explores deviations from true RV. Materials and methods: A custom-built application created in Mathematica was used to explore the effect of altering the humeral orientation on slice angle acquisition by 3D imaging technologies, on the perceived angle of RV from the 2D-projection of the reference axes. The application allows for control of humeral axis orientation relative to image slice (3D) or plain of projection (2D) and humeral rotation. The effect of rotating a virtual model of one humerus around its own axis and in discrete anatomical directions on the measured RV angle was assessed. Results: The coaxial measurement of humeral RV (31.2°) differed from off-axis measurement, with a maximum difference in measured RV of 50° in 45° of extension. The typical position of the humerus in a CT scan resulted in a difference in RV measurement up to 22°. Explorations of deviation led to the following outcomes, as divided by anatomic direction. Extension and abduction led to an underestimation, and flexion and adduction led to an overestimation of the RV-angle. Conclusion: Measurements must be done consistently about the position and orientation of the humerus. Deviation in the humeral alignment of as little as 10° can distort the measurement of version up to 15°.