Postural stability is important in everyday life as falls can cause severe injuries. Risk of injuries is higher in the elderly whose balance is often impaired. Modeling postural stability and the parameters that govern it is important to understand the balance mechanism and allow for the development of fall prevention strategies. Several mathematical models have been proposed to represent postural stability of bipeds. These models differ on the number of degrees-of-freedom (DOF) of the skeletal structure, force generation function for the muscle models, and capability to change their behavior as a function of the task. This work proposes a nonlinear model that captures fall recovery using a hip–ankle strategy. The muscle actuation is modeled as a third-order Poynting–Thomson's (PT) mechanical system where muscles and tendons are represented as lumped parameters actuating the aforementioned joints. Both a regression technique and a Kalman Filter (KF) are used to estimate the muscle–tendon parameters of the model. With a good model, the direct estimation of these parameters would allow clinicians to improve postural stability in the elderly, monitor the deterioration of the physical condition in individuals affected by neuro-degenerative diseases, and develop rehabilitation appropriate processes.