Precise gas turbine engine control for dynamic processes is of great importance in industry. To be able to preciously control the turbine engine an accurate model, which can predict the engine dynamic responses is required. This work focuses on developing a new prediction model for turbine engine rotor speed in a dynamic situation (time-varying rotor speed command). In this new method, the governor model is included, and the influence of governor can be characterized by introducing rotor speed feedback in the model. The model is tested by comparing the long time prediction results with experiment measurements. A close agreement is achieved and the results are significantly more accurate than the model neglecting the governor. We also provide detailed discussion regarding the mathematical form for the reduced model which provides an insight of the governing mechanism. It shows that the governor output can be approximated as multiplication product of the speed error signal and a proportional gain at an operating point.