In the paper, a novel input-constrained H∞ fault-tolerant control approach is developed by using sliding mode control technology and event-triggered reinforcement learning (RL) algorithm. To reduce or even eliminate the impacts of the time-varying actuator failures, a properly sliding mode control strategy is proposed for the controlled system, while the event-triggered H∞ control scheme is established via RL algorithm for the equivalent sliding mode dynamics. By utilizing a single neural network (NN), the Hamilton–Jacobi–Bellman (HJB) equation can be solved approximately, thereby gaining time-triggered worst-case disturbance law, as well as event-triggered optimal control policy. Besides, it is unnecessary to given a initial stabilizing control input in the learning process of neural networks (NNs) in this paper. Moreover, the Lyapunov stability principle is applied to guarantee that the controlled system is uniformly ultimately bounded (UUB). Finally, to verify the feasibility and efficient performance of the developed approach, three simulations are carried out.