A novel event-based decentralized optimal control is designed to handle the modular robot manipulators (MRMs) with health indicator. Firstly, the dynamics of the MRMs is established. Secondly, the developed switching threshold event-triggered mechanism is proposed. Considering the optimal tracking control performance and health status of the system, an improved value function is proposed, which is composed of the system state, control law, fault observations and health state indexes. And the health status indexes are obtained from the health monitoring of the subsystem by utilizing event-triggered health indicator. Then, an event-based critic-only neural network neural dynamic programming (NDP) algorithm is utilized to implement the event-triggered approximate optimal controller design. Finally, the Lyapunov-based stability analysis demonstrates that the MRMs is uniformly ultimately bounded. And the experimental results show that the proposed method performs effectiveness.