The suggestion that the knee behaves like a closed kinematic chain has been tested experimentally. The tibial joint surfaces were replaced by a deformable material and new surfaces were moulded by flexion and extension of the knee. The results support the closed kinematic chain theory; the newly moulded joint surfaces strikingly resemble the original joint surfaces. Inferences are drawn to explain why a fracture of the tibial condyle will remould with early mobilisation of the knee; why a jerk is elicited in tests for anterolateral instability; and why a flexion-compression force alone is unlikely to damage the menisci.