Background Design improvements have increased the success of total ankle replacement, providing patients with end-stage ankle arthritis a viable alternative to arthrodesis. However, revision rates are higher than those for hip and knee arthroplasty, with the most prevalent cause of failure being aseptic loosening. The objective of this study was to quantify and compare the relative bone-implant motion patterns in two well-known total ankle replacement designs. Methods A custom-designed mechanical simulator applied compressive loads (up to 300N) and bending moments (3 Nm) to six pairs of human cadaveric ankles implanted with total ankle replacements, inducing a natural range of motion about three orthogonal axes: plantar flexion-dorsiflexion, inversion-eversion, and internal-external rotation. The implants analyzed were the Agility and the STAR (Scandinavian Total Ankle Replacement). The relative bone-implant motions for each implant component were measured with use of an optical motion capture system. Results The Agility typically exhibited greater relative motion than the STAR, with significant differences for the tibial component in inversion-eversion (p=0.037) and for the talar component in internal-external rotation (p=0.039). The magnitudes of the relative motions were affected by the loading direction and by compression. The motion magnitudes were quite large, with values exceeding 1000μm for the Agility talar component in plantar flexion-dorsiflexion and in inversion-eversion. Conclusions The greater magnitudes of relative motion in the Agility suggest that primary instability of the implant may contribute to its higher clinically observed aseptic loosening rate. Future total ankle replacement designs will require better fixation to improve outcomes. The results underscore the need to conduct preclinical biomechanical assessments of relative motion patterns in ankle replacements. Clinical Relevance Stable initial implant fixation will likely improve clinical outcomes of total ankle replacement.