Aiming to the attitude control system of quad-rotor unmanned aerial vehicles, advanced control scheme should be studied to obtain the satisfied performance. A novel active disturbance rejection generalized predictive control (ADRC-GPC) is presented by combining the technique of active disturbance rejection control (ADRC) and generalized predictive control (GPC). The extended state observer (ESO) of active disturbance rejection control is employed to estimate and compensate the existing uncertainties and disturbance of the nonlinear dynamics systems, such that an integrator form can be obtained to serve as the model for GPC design. By using this scheme, the step response coefficient matrix can be derived analytically and the computational complexity can be substantially reduced. The experiment results show that the designed ADRC-GPC scheme can be applied in the real-time control for the attitude system of the quad-rotor unmanned aerial vehicle (UAV), it can not only meet the need of control accuracy and rapidity, but also have strong disturbance rejection ability and stability. Therefore, the proposed active disturbance rejection generalized predictive control scheme can be used to control under-actuated nonlinear multivariable plants effectively.