This paper studies a novel learning observer-based fault-tolerant tracking control strategy for a hypersonic vehicle attitude system with external disturbance, parameter uncertainty and actuator faults. Without obtaining the prior knowledge of lumped disturbance, a novel learning observer combined with the barrier function is first proposed to estimate the attitude angular velocities and reconstruct lumped disturbance accurately and quickly. By introducing the estimation value into the controller design process, a backstepping fault-tolerant tracking controller is developed to solve the attitude tracking problem of hypersonic vehicle. The stability of the hypersonic vehicle faulty attitude system using the observer-based fault-tolerant control method is analyzed by Lyapunov framework. Finally, the feasibility and superiority of the fault-tolerant control scheme is proved by comparing with different existing observer and control strategy through numerical simulations.