Abstract: With increasingly serious energy shortages and environmental pollution, electric vehicles (EVs) have drawn widespread attention in recent years. The lithium-ion battery is widely used in the field of EVs owing to its superior energy density, life cycle, low self-discharge rate, and maintenance of memory. Prediction of the remaining useful life (RUL) of lithium-ion batteries is a key parameter in battery management systems. The accurate prediction of RUL is a prerequisite to ensuring the safety and reliability of the battery system. The gradual deterioration in the performance of lithium-ion batteries with cycling is normally predicted using capacity and resistance. However, this method is difficult to use in practical applications. To address this problem, a nonlinear autoregressive model with exogenous inputs (NARX) dynamic neural network was proposed to predict RUL. First, according to the discharge data of the lithium-ion battery, three indirect health indicators, namely, cut-off time, constant current time, and peak temperature time in discharge, were proposed, and grey relation analysis (GRA) was used to analyze their relation to capacity. The proposed three indirect health indicators have significant relationships with battery capacity. In addition, due to the influence of temperature vibration, electromagnetic interference, and external disturbance, RUL prediction of the lithium-ion battery is a typical nonlinear problem. In order to cover this weakness, the NARX dynamic neural network was established to predict the RUL of the lithium-ion battery. Finally, a closed-loop and an open-loop NARX were compared with the backpropagation neural network based on particle swarm optimization (BPNN-PSO), least-square support vector machine (LS-SVM), and extreme learning machine (ELM) of existing models under the open data of NASA. The experimental results show that the estimation performance RMSE (NO.5) of the proposed model is improved by about 33% compared with the standard ELM, verifying that the proposed model is superior to other methods in the RUL of lithium-ion batteries.