Previous research investigating the stability of the ankle joint complex may be categorised into two methodological groups, employing either an actuator to perturb the limb, or a form of standing balance disturbance such as a tilting platform, both of which test the joint in single degree of freedom (DOF). The aim of this thesis was to investigate how we control foot position and stabilise the joint when there is potential for movement in three DOF. A secondary aim of the thesis was to model the intrinsic mechanical properties of the ankle joint complex in three dimensions when coupled movement of the tibio-talar and talo-calcaneal joints are possible. This thesis details (i) the development of a perturbation rig that allows foot movement in single- or three-DOF with associated real-time visual target-matching software, and (ii) the use of the rig to investigate the stabilisation of the ankle joint complex in single- and three-DOF. The experimental procedure used a common task performed in three experimental conditions. Subjects were required to maintain a neutral foot position while developing varying levels of plantar-flexion torque. A perturbation was applied to the foot if subjects were within specified tolerance for both foot position and torque, represented by the visual display. Performance of the task in the first condition required the subject to only match torque as the foot position was fixed, with the perturbation being applied in dorsi-flexion (ie, single-DOF). The second experimental condition allowed the foot to move in the sagittal plane, hence subjects were required to control both torque and foot position in single-DOF, with perturbation applied in dorsi-flexion. The third condition enabled movement in dorsi/plantar-flexion, inversion/eversion and adduction/abduction (three-DOF) in both task and perturbation. Subjects were required to maintain the neutral foot position and the necessary torque level. There were three areas of interest common to each experimental protocol. The muscle strategy used to complete the task was investigated using a combination of surface and fine-wire electromyography on lower leg and thigh muscles. The 500ms period prior to perturbation was investigated to determine if synergies were evident between muscles such as medial and lateral gastrocnemius, soleus and peroneus longus. Two classes of activation strategies for the three-DOF condition emerged from the subject population: differential activation of the triceps surae group, and co-contraction. The former strategy may take advantage of the distinct morphology of the lateral gastrocnemius and peroneus longus muscles to best perform the position-matching component of the 3D task. The results suggest that the ankle joint is mostly stabilised in 3D by the intrinsic mechanical actions of the muscles producing plantar flexion moments. The muscles stabilised the foot in inversion, but not in eversion where there was very little motion. However, the different activation strategies employed may have varied efficacy in contributing to joint stability. This form of active stabilisation means that the previous literature focus on reflexes to stabilise the joint may need to be reassessed. Likewise, it may be appropriate to use the perturbation rig to quantify active ankle joint stability in order to assess the probability of ankle injury, rather than the current clinical measures employed. The reflexive response due to the perturbation was examined in the 200ms following perturbation. Variation in the modulation of monosynaptic reflexes was observed between subjects in various muscles in the higher DOF tasks. This is likely due to the differing activation strategies used to perform the task, and the variability in the kinematic response to perturbation. An attempt was made to calculate the intrinsic mechanical properties of the joint in 3-D using the kinematic and kinetic data during the first 15 ms period of perturbation. The system was modelled as a spring-damper using a constrained non-linear least squares, with stiffness and viscous terms for each axis, and inertial tensor elements as variables in the routine. The effect of increased muscle activation on the displacement of the foot about each of the anatomical axes was to significantly lower the movement of the sub-talar joint.