A two-time-scale model-based adaptive fault tolerant control method is proposed to deal with faults of actuator for a helicopter. According to the characteristics of the helicopter in which different states with different settling times, and based on the time-scale separation principle, the helicopter model is divided into the fast model (i.e., the attitude dynamics) and the slow model (i.e., the translational dynamics). The backstepping control and the inverse dynamic control are independently used to design the controllers for the fast model and the slow model in which different control intervals are used. Actuator effectiveness factors (AEFs) are introduced into the two-time-scale model to show the healthy conditions of the actuator. An unscented Kalman filter (UKF) is used for the online estimation of the AEFs. The results from UKF are used for the adaptive reconfiguration of controllers for the fast model and the slow model. Simulation results show that the proposed adaptive fault tolerant control method can eliminate the impact of actuator faults (including constant and time-varying fault) on helicopter. Desirable flight performances and good control effects are achieved.