In the current Ford Pro-Trailer Backup Assist (TBA) system, trailer hitch angle is determined utilizing the reverse camera of the vehicle. In addition to being sensitive to environmental factors such as lighting conditions and occlusion, the vision-based approach is difficult to be applied to gooseneck or fifth wheel trailers. In this paper, a yaw rate based hitch angle observer is proposed as an alternative sensing solution for TBA. Based on the kinematic model of the vehicle-trailer, an instantaneous hitch angle is first derived by utilizing vehicle yaw rate, trailer yaw rate, vehicle velocity and vehicle/trailer parameters provided by the TBA system. Due to signal errors and parameter uncertainties, this instantaneous hitch angle may be noisy, especially at lower vehicle speed. To improve the estimation accuracy, a Kalman filter (or an extended Kalman filter) which is based on the vehicle-trailer kinematic model is then formulated to obtain the optimal estimate, with the instantaneous hitch angle treated as a pseudo-measurement. Experiment results verify the accuracy and robustness of the proposed approach, and that the same algorithm can be applied to both conventional trailers and gooseneck/fifth wheel trailers.