ABSTRACTJoint stability is a primary concern in total knee joint replacement. The GMK Sphere prosthesis was specifically designed to provide medial compartment anterior–posterior (A–P) stability, while permitting rotational freedom of the joint through a flat lateral tibial surface. The objective of this study was to establish the changes in joint kinematics introduced by the GMK Sphere prosthesis during gait activities in comparison to conventional posterior‐stabilized (PS) fixed‐bearing and ultra‐congruent (UC) mobile‐bearing geometries. The A–P translation and internal/external rotation of three cohorts, each with 10 good outcome subjects (2.9 ± 1.6 years postop), with a GMK Sphere, GMK PS or GMK UC implant were analysed throughout complete cycles of gait activities using dynamic videofluoroscopy. The GMK Sphere showed the smallest range of medial compartment A–P translation for level walking, downhill walking, and stair descent (3.6 ± 0.9 mm, 3.1 ± 0.8 mm, 3.9 ± 1.3 mm), followed by the GMK UC (5.7 ± 1.0 mm, 8.0 ± 1.7 mm, 8.7 ± 1.9 mm) and the GMK PS (10.3 ± 2.2 mm, 10.1 ± 2.6 mm, 11.6 ± 1.6 mm) geometries. The GMK Sphere exhibited the largest range of lateral compartment A–P translation (12.1 ± 2.2 mm), and the largest range of tibial internal/external rotation (13.2 ± 2.2°), both during stair descent. This study has shown that the GMK Sphere clearly restricts A–P motion of the medial condyle during gait activities while still allowing a large range of axial rotation. The additional comparison against the conventional GMK PS and UC geometries, not only demonstrates that implant geometry is a key factor in governing tibio‐femoral kinematics, but also that the geometry itself probably plays a more dominant role for joint movement than the type of gait activity. © 2019 The Authors. Journal of Orthopaedic Research® published by Wiley Periodicals, Inc. on behalf of Orthopaedic Research Society. J Orthop Res 37:2337–2347, 2019