A new wearable parallel robot has been designed and constructed for ankle joint rehabilitation treatments. The robot employs four pneumatic muscle actuators (PMA) together with cables to achieve three rotational degrees of freedom (dof) of its end platform. Parallel topology of the robot, unpredictable environment along with the time varying and non-linear behavior of actuators impose modeling and control challenges which are difficult to comprehend. In this paper an optimal fuzzy dynamic model of the pneumatic muscle has been developed to accurately predict the muscle behavior. The model is capable of mapping the complex relationship in length, force and pressure of the PMA with higher accuracy. This model has been further used to develop a fuzzy control scheme for the ankle robot. Experimental results are obtained to study and model the simultaneous actuation of all the actuators. Comparison with the previous dynamic modeling and control schemes demonstrates an improved performance of the proposed fuzzy controller.