Knee forces after total knee replacement are directly related to the transmission of stresses including contact stresses generated at the bearing surface, stresses at the implant-cement-bone interface and stresses transmitted to underlying bone. These stresses are a major factor in wear and fatigue, aseptic loosening and implant migration, bone remodeling, stress shielding and osteoporosis, all of which determine the life of the implant. We report for the first time the in vivo measurement of tibial forces, shear and moments. A custom telemetry-enabled implantable device was used to measure tibiofemoral forces in vivo. We implanted the telemetry-based device into three patients scheduled for total knee arthroplasty. We measured knee forces and kinematics during activities of daily living. The axial component of forces predominated, especially during the stance phase of the activities studied. During walking, forces peaked between 2 and 3 x body weight. Peak tibial forces were substantially higher while climbing stairs (averaging 3 x body weight) than for the chair-rise and squat activities. Overall, shear forces, as well as moments at the tibial tray, were fairly low. We report on the first finite element model of knee joint arthroplasty using in vivo knee forces for the calculation of polyethylene stresses. Walking and stair-climbing generated peak contact stresses below the threshold that is generally considered safe for polyethylene. Contact stresses generated during high flexion activities were substantially higher and largely due to the reduced contact area in deep flexion rather than an increase in contact forces. These results support the use of “high-flexion” designs that improve contact conditions and preserve contact area at high flexion angles. These data can be used to validate existing models of the knee that estimate these forces and to develop more accurate models. Knowledge of in vivo forces can be used to design more effective in vitro knee testing rigs and knee wear simulators that can accurately model knee function and prosthetic wear. Finally, these results can directly enhance prosthetic design.