The purpose of this study was to investigate whether the axis of the humeral articular surface bears a more constant relationship to the bicipital groove than to the distal humerus. This finding could help intraoperative placement of a shoulder prosthesis in the most anatomic position. Magnetic resonance imaging was used in 41 live volunteers and 9 cadavers to measure the distance from the biceps groove to a line drawn perpendicular to the midpoint of the humeral articular surface. A cadaver model with a plastic humeral prosthesis was used to convert this biceps distance into an angular measurement. The relationship between the biceps distance and the retroversion angle of the humeral head with respect to the humeral epicondyles was analyzed. The biceps distance averaged 11.8 mm+/-3.5 mm (equivalent to 26.8 degrees+/-8.3 degrees), whereas retroversion averaged 26.8 degrees +/-12.2 degrees. This difference in variability is significant (P=.008). Placing the fin of a humeral head implant 12 mm posterior to the biceps groove reproduces normal anatomy better than the use of an arbitrary standard of 30 degrees to 40 degrees of retroversion.