Objective: To setup a custom-made gait simulator, and to provide an efficient tool for biomechanics research of ankle and foot. Methods: From November 2017 to April 2018, a total of 6 fresh frozen specimens of the foot and ankle were collected. The donated specimens, free of diseases in the foot and ankle part, were from the Department of Anatomy, Shanghai Medical College of Fudan University. Donors were 3 males and 3 females, aged from 48 to 69 years old, with an average age of 58.8 years old. The nine tendons in the foot were divided into 4 bundles, including anterior group comprised of tibialis anterior (TA), extensor hallucis longus (EHL) and extensor digitorum longus (EDL). Posterior group comprised of Achilles's tendon (AT). Medial group comprised of tibialis posterior (TP), flexor hallucis longus (FHL), and flexor digitorum longus (FDL). Lateral group comprised of peroneus brevis (PB) and peroneus longus (PL). A custom-made gait simulator was set up by using four independent electro motors to actuate 4 bands of tendons in the foot and another six motors to control tibia to achieve 6 degree-freedom parallel mechanism. And a hydraulic machine was used to provide axial pressure along tibia. Gait cycle of six fresh frozen cadaver feet was reproduced using this machine, and the kinematics data of ankle movement and ground reaction force (GRF) data was collected. By comparing the data above with the normal human gait data, the simulation results were analyzed to explore the clinical usage of this machine. Results: On the sagittal plane, the ankle appeared to plantarflexion at the beginning of gait, and then turned to dorsiflexion after the max plantar flexion (about 10°) at 18% of gait cycle. At the 40% gait cycle, ankle joint was in neutral position and reached its max dorsiflexion (about 22°) at 83% gait cycle. On the coronal plane, ankle joint appeared inversion at the beginning and eversion afterwards with 10° range of change. On the horizontal plane, movement of ankle joint was small. Results showed that the first peak of vertical ground reaction force can reach to 1.1-1.3 times of bodyweight at 25% of gait cycle and the second peak appeared little lower at 70% of gait cycle. The GRF in posterior direction reached its peak at 30%, and then turned to anterior with its peak at 83% gait cycle. The GRF had small variation at the internal to external direction.All of the coefficients of multiple correlations (CMC) of GRF and ankle joint movements were close to or greater than 0.90. Conclusion This custom-made gait simulator has good gait simulation ability with high intra repeatability in respect of ankle rotation and ground reaction force, and can satisfy the request for ankle and foot biomechanics research.