Abstract: The state of charge (SOC) of lithium batteries is an important performance indicator in an electric vehicle's battery management system (BMS). However, online estimation of SOC is difficult due to technological limitations. We propose a SOC estimation method for lithium battery based on kernel extreme learning machine (KELM) optimized by improved grey wolf algorithm (IGWO). Aiming to reduce the search time of the grey wolf optimization algorithm (GWO), we utilize the chaotic tent mapping and opposition-based learning to initialize individual position and then integrate the nonlinear decline of convergence factor to improve the search ability. Finally, we employ the Gaussian mutation operator and greedy strategy to update the global optimum of each generation. The IGWO is applied to optimize the parameters of the KELM model, which is then employed to estimate the SOC of the lithium battery. Several experiments are conducted based on the measuring data from the ADVISOR operation mode and the constant discharge test for the lithium battery, where the results show that the estimation effect of the proposed model is better than that of ELM, KELM, GWO-KELM, EKF, and unscented Kalman filter. Hence, the proposed method has a good guiding significance for a BMS.