INTRODUCTION. Useful feedback from a Total Knee Replacement (TKR) can be obtained from post-surgery in-vivo assessments. Dynamic Fluoroscopy and 3D model registration using the method of Banks and Hodge (1996) [1] can be used to measure TKR kinematics to within 1° of rotation and 0.5mm of translation, determine tibio-femoral contact locations and centre of rotation. This procedure also provides an accurate way of quantifying natural knee kinematics and involves registering 3D implant or bone models to a series of 2D fluoroscopic images of a dynamic movement. AIM. The aim of this study was to implement a methodology employing the registration methods of Banks and Hodge (1996) [1] to assess the function of different TKR design types and gain a greater understanding of non-pathological (NP) knee biomechanics. METHODS. Knee function was assessed for five subjects with NP knees (4 males and 1 female, 34.8 ± 10.28 years, BMI 25.59 ± 3.35 Kg/m. 2. ) and five subjects 13.2 (± 1.8) months following a TKR (2 males, 3 females, 68 ± 9.86 years, BMI 30 ± 3 Kg/m. 2. ). The TKR types studied included 1 cruciate retaining, 2 cruciate substituting, 1 mobile-bearing (high flex) and 1 medial pivot). Ethical approval was obtained from the South East Wales Local Research Ethics Committee. Each subject's knee was recorded whilst they performed a step up/down task, using dynamic fluoroscopy (Philips). 3D CAD models of each TKR were obtained for the TKR subjects. 3D bone models of the knee, tibia and femur were created for the 5 NP subjects by segmenting MRI scans (3T GE scanner, General Electric Company) using ScanIP (Simpleware, Ltd.). Using the program KneeTrack (S A Banks, USA), each TKR component and bone model was projected onto a series of fluoroscopic images and their 3D pose iteratively adjusted to match the contours on each image. Joint Kinematics were determined from the 3D pose of each 3D model using Cardan/Euler angles [2]. The contact points and centre of rotation of each TKR were also computed. RESULTS. The mean range of motion (ROM) in the sagittal plane was 61° for the NP cohort and 64° for the TKR cohort. The mean frontal plane ROM was 4° for NP knees and 3° for TKR. A greater axial ROM was achieved by the mobile-bearing (7.5°) and medial pivot TKR (7.0°), compared to the cruciate retaining (4.4°) and substituting (3.6°). The Medial Pivot TKR rotated around a medial centre of rotation, whereas the centre of rotation was located laterally for the other TKR types. This has also been found in other studies of stair climbing activities [3]. CONCLUSIONS. This study demonstrates how this method can be used to quantify and compare the kinematics, contact locations and centre of rotation for a range of TKR designs and NP knees in-vivo. Initial analyses have identified functional differences associated with different TKR designs