Nineteen fresh-frozen anatomic specimens of the ankle were mounted in an experimental test device with the foot in supination. Torque versus rotation curves were recorded as each foot was rotated externally to failure. The mean failure torque was 45.3 Nm, the mean rotation to failure was 41.4 degrees, and the mean energy absorbed to failure was 10.6 Nm. Twelve specimens failed by fracture of the fibula, five by failure of lateral ligaments with no fibular fracture, one by a subtalar dislocation, and one by fracture of the calcaneous through a fixation screw hole. There were no statistically significant differences in torque or energy to failure between those specimens that failed by fibular fracture and those with ligamentous failure alone. Rotation to failure was 6 degrees greater in the ligamentous failure group. Ankles with less initial rotatory laxity tended to fail at higher torque levels. The mean ankle fracture torque measured in this study is similar to previously reported mean ligamentous failure torque for the knee flexed to 20 degrees and rotated externally to failure. In a prior in vivo study, approximately 10 Nm of external foot torque and 20 degrees of foot rotation could be tolerated prior to the initiation of pain. The present study indicates that these in vivo pain threshold levels represent approximately 50% of the rotation to failure and 25% of the torque to failure.