The aim of this study was to quantify the influence of prosthetic humeral head shape, as well as the number of available prosthetic head sizes, on replicating the normal humeral head anatomy during shoulder reconstructive surgery. Computer modeling software was used to create virtual sets of both spherical and elliptical prosthetic heads, which were virtually implanted into 3-dimensional computed tomography scan-based models of 79 proximal humeri. Anatomic replication was considered successful if the measured parameters (diameters of the base of the head in the frontal and sagittal planes, radii of curvature in the frontal and sagittal planes, and humeral head height) were all reproduced within 3 mm. The Fisher exact test was used to compare the percentage of successful replications for both head types and to compare differences resulting from the use of sets with fewer or more available head sizes. Statistical significance was set at P ≤ .05. Regardless of the number of available head sizes per set, it was possible to replicate the normal anatomy within 3 mm in a higher percentage of specimens using elliptical (96%-100%) as opposed to spherical (41%-78%) prosthetic heads (P ≤ .0013). Compared with use of spherical prosthetic heads, use of elliptical heads resulted in improved replication of the normal humeral head shape. In light of the emerging evidence that use of anatomically shaped prosthetic humeral heads might lead to better shoulder function and possibly improved implant survivorship, the findings of this study may have important clinical and economic implications.