[Objective] To explore the effect of different configurations of cannulated screws on the stability of ankle joint fusion by fi-nite element analysis. [Methods] Thin-layer scan from knee joint to sole was performed in a 37-year-old male volunteer by spiral CT, andDICOM images were obtained and imported into Mimics to segment three-dimensional point clouds of calcaneus, talus, tibia and fibula,and then, the finite element model of ankle joint was simulated in Geomagic and Hypermesh. The normal control model and the ankle jointfusion models with double screws parallel (DSP), double screws crossover (DSC) and triple screws crossover (TSC) were established respec-tively. The relative displacement of the tibial talar joint surface, the Mises stress peak value and the mean Mises stress in the model were ob-served. [Results] The DSP group proved significantly greater than the DSC and TSC groups in term of displacements under loading, includ-ing plantar flexion [(106.2±8.6) μm vs (58.9±4.5) μm vs (58.5±4.1) μm, P＜0.001] internal rotation [(101.6±6.4) μm vs (59.1±4.4) μm vs(58.5±4.0) μm, P＜0.001] and external rotation [(101.5±6.6) μm vs (58.5±4.6) μm vs (58.3±4.2) μm, P＜0.001], while there was no signifi-cant difference between the DSC and TSC groups (P＞0.05). The DSP had significantly greater peaks Mises stress on the implant than DSCgroup and the TSC group (65.5 MPa, 51.3 MPa, 50.5 MPa). In addition, the Mises stress on the tibiotalar articular surface in the DSP groupwas significantly less than those in the DSC and the TSC groups (P＜0.05), whereas which was not statistically significant between DSCgroup and TSC group (P＞0.05). [Conclusion] The DSC and TSC have significantly better stability of the screw placement configuration thanthat in DCP placement, while good stability of the screw palcement is conducive to improving the success rate of arthroscopic ankle arthrod-esis.