Aims: Most mobile-bearing knee prostheses feature bearing rotation; the benefit of also providing AP translation remains controversial. We have compared the kinematic behaviour of the CERAGYR mobile-bearing knee with two different polyethylene bearings, one with a pin-on-hole mechanism allowing only rotation, the other with a pin-on-slot also allowing 5 mm AP translation of the bearing. Methods: A computer-aided study was made on lateral knee X-rays in extension and in 90° flexion, 12 to 36 months following arthroplasty. 30 knees in each group were studied under non-weight bearing conditions (NWB) and 16 under weight-bearing conditions (WB) during a stepup test. Bearing mobility was calculated based on measured displacements between landmarks on the tibial baseplate and in the PE bearing. Results: No AP translation was noted during flexion in the pin-on-hole group; posterior translation of the base-plate during flexion was noted in the pin-on-slot group (mean values: 1.66 mm NWB, 1.37 mm WB). Bearing rotation during flexion was greater in the pin-on-slot group than in the pin-on-hole group (mean values: 6.5° vs 4.4° WB; 3.8°vs 3.3° NWB). A screw home mechanism was noted in 14/16 knees under weight-bearing conditions in the pin-on-hole group, and in 7/16 in the pin-on-slot group. Clinical evaluation showed no difference in outcome between the two groups. Conclusions: Based on these findings, the kinematics of the CERA-GYR knee prosthesis with a pin-on-hole mobile bearing appear closer to normal than with a pin-on-slot bearing, as the latter was found to exhibit reverse femoral rollback as well as reverse screw home in the majority of cases.