The human proximal tibiofibular joint (PTFJ) and its relationship to overall knee joint mechanics have been largely unexplored. This study describes force/displacement data from experiments done on four human cadaveric knee specimens and general conclusions obtained with the help of a statistical modeling technique. Specimens were rigidly affixed at the tibia to a force plate and the femur was attached to a custom made device allowing for manual load application. Motion of the fibular head was tracked relative to the tibial plateau by means of reflective markers and a high speed digital camera synchronized with the force plate data stream. Each specimen was subjected to a range of loading conditions and a quadratic regression model was created and then used to predict the specimen's response to standardized loading conditions and compare these across specimens. Statistical analysis was performed with a three-factor analysis of variance with repeated measures. Proximal tibiofibular joint motion was largest in the anterior-posterior direction with translations of 1-3 mm observed during a range of physiological loading conditions. The applied internal-external rotation moment had a significant effect on proximal tibiofibular joint translation (P < 0.05). Effects of varus-valgus loading and flexion angle were seen in some specimens. This study demonstrates that substantial proximal tibiofibular joint motion can occur in physiologic loading states. Preservation of proximal tibiofibular joint function, and anatomical variations which affect this function, may need to be considered when designing surgical procedures for the knee joint.