The mechanisms of knee fracture were studied experimentally using cadaveric knees and analytically by computer simulation. Ten 90 degree flexed knees were impacted frontally by a 20 kg pendulum with a rigid surface, a 450 psi (3.103 MPa) crush strength and a 100 psi (0.689 MPa) crush strength aluminum honeycomb padding and a 50 psi (0.345 MPa) crush strength paper honeycomb padding at a velocity of about five m/s. During rigid surface impact, a patella fracture and a split condylar fracture were observed. The split condylar fracture was generated by the patella pushing the condyles apart, based on a finite element model using the maximum principal stress as the injury criterion. In the case of the 450 psi aluminum honeycomb padding, the split condylar fracture still occurred, but no patella fractures were observed because the honeycomb provided a more uniform distribution of patella load. No bony fractures in the knee area occurred for impacts with a 50 psi paper honeycomb padding. In the four 100 psi aluminum honeycomb tests, there were two split condylar fractures and no bony fractures in the other two tests.A 3-D finite element human knee model was developed to try to understand the internal stress distribution during knee impacts. The model included a femur, patella, tibia, associated ligaments modeled as bar elements, and a pendulum. The calculated pendulum force for a rigid impact and impacts using the 100 psi and 450 psi honeycomb matched that obtained from the tests but for the 50 psi honeycomb case, it did not. The calculated stress concentration location in the femur coincided clearly with the fracture location in the rigid impact test. To protect the knee, the ultimate crush strength of the honeycomb padding was presumed to be 90 psi (0.620 MPa), based on linear interpolation of this limited series of experiments.