In this paper, an event-triggered neural network extended state observer is designed for Autonomous Underwater Vehicle (AUV) with system uncertainties, external disturbances and limited communication. First, the signals transmission in sensor-to-observer channel depends on the established event-triggered mechanism to reduce communication resources. Then, the system uncertainties and external disturbances are estimated online by utilizing a neural network (NN) estimator, which can improve the estimation accuracy of the observer, and the weights of the NN estimator are updated only when the designed event is triggered, which can significantly reduce the computation burden. It is proved that the estimation error is bounded, and Zeno behavior is excluded. Finally, some simulations are demonstrated to verify the effectiveness of the proposed observer.