This article presents an EMG-to-moment optimization model suitable for clinical studies to estimate the contribution of agonist and antagonist muscle groups to the net ankle joint moment during dynamic and isometric tasks. The proposed EMG-to-moment model took into account realistic muscle properties such as the electromechanical delay, and a force-length-velocity relationship with subject-specific muscle anthropometric data. Subjects performed isometric ankle plantar-flexion (fixed-end contraction) and dynamic tasks (heel-raise) in two positions, seated and upright. Two models were compared: the proposed EMG-to-moment model calibrated on eight dynamic and isometric tasks (Model 8-tasks) and on two dynamic tasks (Model 2-tasks), and a published reference model. First, each model was calibrated at the ankle joint on 10 subjects by adjusting individual set of parameters to estimate the muscle groups contributions. Then, the model was used to predict the ankle net joint moment. The model developed in this study showed good prediction. The Model 8-tasks predicted net joint moment with an average RMS error of 6.11 +/- A 4.41 N m and a mean R (2) of 0.67 +/- A 0.26 across dynamic and isometric tasks. The proposed EMG-to-moment model was simple and required few calibration tasks without oversimplifying muscle properties, satisfying requirements for clinical settings.