Compared with fixed-wing aircraft, small-size unmanned helicopter is more superior in maneuverability and flexibility. It can be widely used in the military and civilian fields. In recent years, many researchers developed unmanned helicopter flight control system. As we cannot obtain the objective without getting its dynamic model, the subject of helicopter modeling has drawn much attention. This thesis focuses on the modeling of the Thunder Tiger-60 small-size unmanned helicopter. It gives a brief introduction on the dynamic model of the main rotor, tail fins and horizontal tail, and obtains linear model of the helicopter at its operating points at hover. The mathematic model of helicopter is a MIMO state equation. We study the model identification in single separated degree of freedom, including of pitch, roll, yaw and vertical channels. In order to obtain identification data, this thesis designs a data acquisition system, which employs ARM9, inertial navigation equipment, GPS, wireless transmitter module and camera sensor. Through Linux operating system, the helicopter flight data is collected and stored, such as PWM signal and body attitude information. Based on the flight data, pitch, roll, yaw and vertical channel models are identified by the least-squares method. The results of the model validation show that the separated channel model identified can predict the helicopter corresponding channel's dynamic performance at hover.