Purpose of the study: In vivo kinematics of the knee joint (anteroposterior translation or rollback, axial rotation, elevation of the femoral condyle, range of motion) was determined for the knees of 30 subjects with a total knee prosthesis with a fixed or mobile plateau and also for the normal knees.

Material and methods: Videofluoroscopic images were recorded during gait and maximal flexion. An automatic 3D adaptation-modelling process was then applied to the fluoroscopic images to determine knee kinematics.

Results: For the normal knee, a certain degree of femoral rollback was noted for the lateral compartment (4.2 mm on average) while minimal translation was observed medially. The femoral rollback increased laterally during maximal flexion (14.4 mm on average) while the medial translation was minimal (1.5 mm on average). Thus, the average movement, which was not observed for all normal knees tested, was a pivot movement centred medially. The variability observed during maximal flexion was wide for all knee prostheses with a fixed or mobile plateau which do not have a stabilising system substituting for the absent posterior cruciate. During flexion, the normal knees exhibited mean 10° external rotation of the over the tibia. All of the rotational knee prostheses presented external rotation (mean 5°, ragne 0–10°). Inversely, the posterostabilised prostheses exhibited medial rotation of the femur over the tibia (mean 5°, range 0–10°), i.e. paradoxical movement.

Discussion: Unlike the normal knee where femoral rollback occurs during maximal flexion, paradoxical anterior translation of the femorotibial point of contact after arthroplasty, in particular in subjects with a fixed plateau prosthesis. For prostheses substituting for the posterior cruciate, femoral rollback involving the lateral condyle occurs regularly with minimal variability in the femorotibial contact point due to the regular engagement of the cam and cam follower mechanism during maximal flexion.