Aims: The aim of the study was to determine the accuracy of the kinematical determination of the centre of rotation of the hip and to compare the outcome of the Computer assisted surgery (CAS) group versus a control group of patients with conventionally instrumented TKA, in a prospective randomized way. Methods: A prospective, randomized and controlled trial was undertaken with an image based CAS system (ION®), using specific knee software for the GENESIS II®total knee system. Randomization was performed on a consecutive group of 50 primary TKA’s, without exclusion criteria. All computed kinematical centres of rotation of the hip were compared to the anatomic fluoroscopic images. The difference between the kinematical centre of rotation and the anatomic centre of the femoral head was measured in the frontal plane. Coronal alignment was measured on full leg standing films. Validation of the full leg standing films was carried out in comparing the pre-operative measured angle and the computed deformity angle at the beginning of the surgery. Outcome of the CAS group was compared to the conventional group on the following items: tourniquet time, operative time, blood loss, patellar alignment, tibial slope, coronal alignment, range of motion and complications. Results: ACCURACY: The correlation index between pre-op full legs and CAS measured values was excellent: r²=0.997. Difference between kinematical centre of rotation and anatomic centre of the hip: mean deviation between the two points was 1.2 mm (0–4mm), stdv 1.2 mm. This corresponds with a mean angular deviation of 0.17° (0–0.57°). OUTCOME: Tourniquet time: conventional 56 min., CAS 72 min. p=0.002. Operative time: conventional 70 min., CAS 93 min. p< 0.001. Blood loss: conventional 3.3 g/dl, CAS 4 g/dl. Patellar alignment: no tilt > 5°, no subluxation > 3 mm, both groups. Tibial slope: conventional 3.5°, CAS 3°. Post-operative mechanical alignment was between 0 and 2° of deformity for 16 conventional knees, and between 3–4° for 5 conventional knees. In the CAS group, all 21 knees scored between 0° and 2° of mechanical alignment. ROM at 6 weeks: flexion conventional 106°, CAS 105°. Fixed flexion contracture: conventional 2.9, CAS 2.1. Complications: delayed wound healing: conventional 2, CAS 1. Conclusions: Computer assisted kinematical determination of the centre of the hip can be highly accurate. Post-operative coronal alignment in CAS group is excellent, however not significantly better than conventional instrumentation.