The anterior portion of the anatomical neck is used as a reference for the osteotomy in shoulder arthroplasty. Resection at this level is thought to remove a segment of a sphere which can accurately be replaced with a prosthetic implant. The objective of the study was to analyse the cartilage/metaphyseal interface relative to an ideal osteotomy plane to define points of reference the may recover retroversion accurately. Data were collected from 24 humeri using a novel technique, combining data acquired using a Microscribe digitiser and surface laser scanner. Rhinocerus NURBS modelling software was used to analyse the Cartilage/metaphyseal interface. The retroversion angle was calculated for the normal geometry and for the standard osteotomy along the anterior cartilage/metaphyseal interface. An ideal osteotomy plane was then created for each specimen and the perpendicular distance from the cartilage/metaphyseal interface was determined, identifying points of least deviation. The reference points were used to simulate a new osteotomy for which retroversion was calculated. Paired t-tests were used to compare the novel osteotomy and traditional osteotomy to the normal geometry. The mean retroversion for the normal geometry was 18.5±9.0 degrees. The mean retroversion for the traditional osteotomy technique was 29.5±10.7 degrees, significantly different from the original (p< 0.001). The mean retroversion using the novel osteotomy was 18.9±8.9 degrees and similar to the normal geometry (p=0.528). The traditional osteotomy resulted in a mean increase in retroversion of 38%. The increase in version may result in eccentric load on the glenoid, an alteration to the rotator cuff balance and poor clinical outcome. The novel osteotomy based on points identified around the cartilage/metaphyseal interface that deviated least from an ideal osteotomy plane resulted in more accurate recovery of head geometry. The novel technique may improve clinical outcome. Further investigation is warranted.