For electric vehicles (EVs) and hybrid electric vehicles (HEVs), the accurate estimation of state of charge (SoC) has always been the focus of research. The commonly used SoC estimation methods are based on equivalent circuit model (ECM) and require high model accuracy. With battery aging, its impedance and capacity will change, which inevitably affects the estimation accuracy. In this paper, the impedance spectrum detection method is integrated into the battery management system (BMS), and a new model updating strategy based on electrochemical impedance spectroscopy (EIS) is proposed for SoC algorithm. Firstly, the evolution of impedance spectrum with battery aging is analyzed through accelerated aging test. It is found that the aging mainly causes the increase of ohmic resistance for the tested Li-ion battery but has no significant impact on the charge transfer and diffusion processes. Therefore, the model parameters and capacity are modified according to the change rate of ohmic impedance. Secondly, the proposed method is verified under charging and discharging conditions. Results show that as the battery ages, the SoC estimation value will be lower than the real value under the discharging condition and higher than the real value under the charging condition. Meanwhile, the maximum absolute error (MaxAE) will reach 29.9 % after 1000 cycles. In contrast, based on EIS correction, the MaxAE can be controlled within 5.4 %, which effectively enhances the estimation accuracy.