Disclosed is a method and system of gesture recognition in a touch display device, which is able to predetermine gesture inputs possibly to be made by a user prior to the completion of the user's touch input, and enable a display unit to display all possible similar gesture inputs so as to provide an instruction (or navigation guidance) for the user. Thus, when using a large-sized touch display device, the user does not have to perform touch operations widely throughout the screen of the display device, because the system can recognize the similar gesture inputs in advance, which renders it easier for the user to operate on the touch display device, thereby obtaining a better user experience.