Unicompartmental knee arthroplasty (UKA) gradually becomes a promising alternative treatment to high tibial osteotomy in patients with unicondylar involvement, in addition to the total knee arthroplasty. In lots of previous series, osteolysis and aseptic loosening were continued to plague surgeons. Insert wearing is considered one of the major causes. The purpose of this study intends to clarify the main relations of the insert wearing to factors like implant designs, alignment of components, and contact forces. In the present study, the finite element analysis (FEA) was used to study the stress distributions of polyethylene (PE) component and to explore the possible wear mechanism. In order to evaluate the wear behavior of a tibial insert, three-dimensional FE models of different designs of unicompartmental knee prostheses (UKP) are constructed to investigate the effects of mal-resection including varus/valgus tilts, and posterior slopes, PE properties as well as contact forces applying on it. Through the study, all material properties of the components are set as close those currently used as possible. And, a 1435 N compression load is applied from the top of medial unicondyle knee at zero degree of flexion. In the correction of varus deformity, the tibial insert has the highest stress at the 15∘varus tilt. In addition, the stress demonstrates no significant difference in response to changes of posterior slopes. Or, the maximal stresses occur at the center of the contacts and are located ca. 2 mm beneath the articulating surface. This result implies that the failure mechanism of the tibial insert initially takes place inside the PE, and then gradually propagates into cracks or flakes from the inside to the contact surface. Furthermore, the maximal stress of the tibial insert shifts laterally with the increasing varus angles. Regardless of varus tilts, the stresses always exceed the yield strength of PE under the gait loading condition. In cases of mal-resection on the tibial plateau in the coronal plane, it was found that anatomically designed unicompartmental knee prosthesis (UKP) can allow more positional errors in varus tilts than that of a symmetrically designed one. However, both should avoid any positional error greater than 5° varus/valgus tilts. Otherwise, it may result in severe wear on its PE component. Since there exist several choices of the PE thickness, the present study also investigates the possibility of increasing thickness to reduce the maximal stress. However, the results of in vivo FE simulations have revealed that a thicker PE insert may not be a good alternative in the sense of reducing PE wear. In fact, a thicker PE component makes almost no significant reduction to the stress of all insert samples. The PE stress increasing strongly depends on the load that applying onto it, instead of its thickness. As a conclusion, the present study would suggest that loading on the medial unicondyle should avoid greater than ca. 800 N. Or, the life of the PE insert may be jeopardized. This threshold value of 800 N has been further verified by the in vitro experiment. In other words, the study suggests that the UKA is more adequate for those of body weights less than ca. 56 kg. In order to give this conclusion more solid, the present report also provide a simply mathematical model for surgeons to select appropriate patients so that a better UKA outcome can be achieved. Overall, the results of present study have shown that the wear of PE tibial components have considerable impacts to UKP, such as joint overuse, overweight, under-/over- corrections of varus deformity and tibial mal-resection. However, the influences of PE thickness and the posterior slopes are minor and can be ignored. For UKA to have excellent long-term results, the body weight, the alignment of components, and accurate technique of surgery will all be required. Nevertheless, under the current circumstances, UKA is not recommended for the over-weighted, young patients or those with excessive activities. Meanwhile, the constraint given from the present study can be a good guild line for the selection of patients.