In this paper, the system identification to determine the mathematical parameters of a plant based on observations on the plant is proposed. For a practical mechanical system, some parameters are too complicated to obtain from theoretical derivation. The system identification is a useful technique to reconstruct dynamic model for helicopters. There is variety of system identification algorithm suitable for different kinds of problem. In this paper, the Levenberg Marquardt (LM) method is selected to identify the unmanned aerial vehicle dynamic. This methodology is then applied to the flight experiment data. The ultimate goal of this project is to extend the current work to build a fully functional autonomous rotorcraft. Numerical simulations are given to demonstrate the validity of our results.