The role of the posterior cruciate ligament (PCL) after total knee arthroplasty has been controversial. Previous studies have reported that function of the preserved PCL after TKA was questionable and that it was difficult to determine the appropriate PCL tension to reproduce rollback. However, several in vivo studies have reported that prosthesis geometry directly affects knee kinematics, making it difficult to determine which factors most influence knee kinematics. The purpose of this two-center, two surgeons study was to evaluate knee kinematics of a single design of CR TKA. A total of 23 TKAs were studied fluoroscopically during a single-limb step-up/down maneuver. The average patient age at the time of TKA, knee score (HSS/KSS) and ROM were 70.6 years, 91.1 points and 116.9 degrees respectively. All patients had a PCL-retaining prosthesis of the same design using an unconstrained “flat” tibial insert. TKAs were performed by one surgeon at each hospital (Group 1:13 knees, Group 2: 10 knees). Both groups of knees exhibited ‘screw-home’ type axial rotations from 20° of flexion to full extension. In Group 1, rollback occurred early in the flexion range and was maintained until 80° of flexion. In Group 2, the lateral condyle exhibited rollback in early flexion, but both condyles translated forward as flexion increased to 80°. Medial and lateral contact were more posterior in Group 1 over most of the range of motion (p< 0.05). Although femoral rollback has been infrequently observed in similar studies of PCL retaining arthroplasties, our two-center, two surgeon data suggest that rollback can be achieved using this unconstrained prosthesis with PCL retention. However, there were consistent and statistically significant differences in the knee kinematics exhibited by the two groups of patients.