Osteoarthritis (OA) is the most common form of arthritis and it affects 27 million US adults. OA disease involves all of the tissues of the diarthrodial joint and ultimately, may lead to softening, ulceration, loss of articular cartilage, sclerosis and polished appearance of the subchondral bone, osteophytes, and subchondral cysts. The first metatarsophalangeal joint (MTPJ1) is affected in up to 42% cases of OA. Besides osteoarthritis, other conditions such as rheumatoid arthritis and gout also affect the MTPJ1. Involvement of MTPJ1 with these conditions invariably leads to deformed toe such as hallux valgus and hallux rigidus. Over 150 surgical techniques exist for treatment of hallux deformity, which includes cheilectomy, arthrodesis, osteotomy, resection arthroplasty, and replacement of part or the entire articular surface with an implant. A hemi-implant, which partially replaces the 1st metatarsal head with minimal bone resection and without altering the sesamoid articulation has shown promising results and gives superior postoperative range of motion and pain reductions. But the geometry of such implants has not been explained in any literature and there are no details of the data used for designing such implants. An anatomically based approach to design the geometry of an MTPJ1 implant is needed in order to best fit the articulating surface of the adjacent phalanx. In the current study, a method was developed for designing a hemiarthroplasty implant for MTPJ1 based upon the morphology of metatarsal. Ninety-seven metatarsal osteological specimens were scanned using a laser scanner to obtain 3D surface data. After aligning the surface data, the articular surface of each metatarsal head (MTH1) superior to the inter-condylar ridge were characterized by a section of an ellipsoid using non-linear unconstrained optimization (NLUO) and the section of the ellipsoid forms the surface of the implant. The implants based upon osteological specimens had a very good fit to metatarsal articulating surface with root mean square error of fit in the range of 0.29 and 0.42mm.The cartilage region, in 14T MRI image from 1st metatarsal of two cadaver feet, was segmented semi-automatically, and a three-dimensional surface of the cartilage shell was created. The average thickness profile of the cartilage on articular part of MTH1 was obtained. For articulating surface of individual osteological surface data, a surface which simulates cartilage outline was created using the cartilage thickness profile. This cartilage outline surface was again characterized with the best fit ellipsoid using NLUO. The parameters of ellipsoid for the cartilage outline surface and the osteological surface were compared. Although the difference between the parameters for the ellipsoid obtained in these two conditions were not found to be significant (p=0.05), this result needs to be validated with more cartilage samples. In addition, scaling for size of the bones should be considered in calculation of the thickness of cartilage.Thus, a new method to design the implant for MTH1 for arthroplasty was identified based upon bones from general population using numerical technique. This method can be extended in designing implants for other joints which need hemi-arthroplasty.