A major challenge in human movement analysis is the development of a non-invasive method to determine individual muscle forces contributing to overall joint actuation. The objective of this research was to develop an electromyographic (EMG)-driven model to predict knee joint dynamics, and validate the model with data incorporating a physiological change to the system. The model proposed differs from other models in that it takes into account varying fibre-type distribution, and incorporates a natural activation recruitment scheme and a non-linear EMG-to-muscle activation function. Results of this single-subject study revealed that the model predicted dynamic responses of the knee joint with a correlation value of r = 0.89 and mean relative error = 15.8%, and a correlation of r = 0.83 and relative error = 20.8% when a physiological change was introduced. The model could potentially be used as a predictive tool for muscle dynamic responses.