This paper investigates the fault-tolerant containment control problem multi-agent systems. Most existing results on fault-tolerant control considered leaderless systems or systems with only one leader. In this paper, an adaptive fault-tolerant control protocol based on adjacency error is proposed to address actuator failures in systems with multiple leaders and multiple followers. This control protocol can compensate for actuator faults and nonlinear disturbances of each follower, and allow the states of all follower to converge to the convex hull formed by leaders. The Lyapunov method is utilized to verify the theoretical results. Finally, a simulation example is given to demonstrate the effectiveness of the fault-tolerant control protocol.