The Re-Link Trainer (RLT) is a novel end-effector device designed for gait retraining post-stroke. The user's foot is constrained to a specific kinematic trajectory relative to the trainer, while the hip and knee are unconstrained. As the RLT only fixes the footplate trajectory the expected constraint on the hip and knee angles will be subject-specific due to individual lower limb geometries. This study had two objectives (1) to calculate the subject-specific theoretical joint angle trajectories the RLT should constrain the hip and knee angle to using computer simulation, assuming a fixed hip position relative to the RLT, and (2) experimentally determine the actual hip and knee joint angle trajectories of healthy users walking in the RLT, and compare them to the theoretical joint angle trajectories. The root mean square error between joint trajectories obtained from motion capture and simulation ranged from 4.31o to 20.51o for the hip and between 4.48o to 22.58o for the knee, suggestive of moderate to poor accuracy and distinct kinematic adaptation strategies when using the RLT. A linear fit method (LFM) was used to determine the similarity between the obtained and simulated joint angle trajectories. LFM results would suggest that users' hip and knee joint angles follow the simulated joint angle trajectories when walking in the RLT, however the actual joint angle trajectories are offset from the simulation trajectories. Post hoc analyses suggest hip motion influences the hip and knee angle trajectory differences for participants.