The object of the present study was to investigate the three-dimensional kinematic and dynamic behaviour of the human ankle joint during walking and in particular to calculate joint and muscle forces. A review of the relevant literature is presented. In the analysis, the ankle is treated as consisting of two joints: the talocrural (Tc.) and the Talocalcaneonavicular (Tcn.). Five cadaveric legs were dissected to ascertain the joint axes, the lines of action of the relevant tendons, and the positions of the ligaments. Seven adult male subjects were studied during barefoot level locomotion, and on planes sloping sideways at + 10 . Observations were made with a force plate, and with three orthogonally placed cine-cameras. Stance phase forces were estimated as follows: inertia forces were neglected; anthropometric data from the cadavers were scaled to fit the walking subjects; the number of unknown forces was reduced to match the available equations by (a) combining muscles into four groups, and (b) assuming no antagonistic activity. Two free bodies were used: the hindfoot plus the talus alone. The Mark 1 solution excludes the Peroneal and Posterior Tibial groups; the Mark 11 solution includes them. The Mark 11 model gave a peak resultant Tc. joint force of 3.9 Body Weight (B.W.) mean, for normal locomotion. The Tcn. joint anterior and posterior facet peak resultants are 2.4 and 2.8 B.W. respectively. The Anterior Tibial/Calf group peak forces are 1.0/2.5 B.W. mean. The Peroneal/Posterior Tibial group peak forces are 0.7/1.1 B.W. mean. Side slope walking is shown to require greater Peroneal/Posterior Tibial muscle group activity than in normal locomotion. A Mark III model, including ligamentous constraint, is presented and recommendations made for its development. The relationship between the results of the study and clinical problems such as joint replacement is considered.