Despite their well-known limitations, in vitro experiments have several benefits over in vivo techniques when exploring foot biomechanics under conditions characteristic of gait. In this study, we present a new setup for dynamic in vitro gait simulation that integrates a numerical model for generating the tibial kinematics control input, and we present an innovative methodology to measure full three-dimensional joint kinematics during gait simulations. The gait simulator applies forces to the tendons. Tibial kinematics in the sagittal plane is controlled using a numerical model that takes into account foot morphology. The methodology is validated by comparing joint rotations measured during gait simulation with those measured in vivo. In addition, reliability and accuracy of the control system as well as simulation input and output repeatability are quantified. The results reflect good control performance and repeatability of the control inputs, vertical ground reaction force, center of pressure displacement, and joint rotations and translations. In addition, there is a good correspondence to in vivo kinematics for most patterns of motion at the ankle, subtalar, and Chopart's joints. Therefore, these results show the relevance and validity of including specimen-specific information for defining the control inputs.