The invention discloses a method for estimating residual capacity of batteries in battery management systems, and belongs to methods for estimating residual capacity of batteries. The method includessimultaneously referring to Extended Kalman Filter (EKF) algorithms and ampere-hour integral processes; processing results S1 of the EKF algorithms and results S2 of the ampere-hour integral processesto obtain ultimate estimated values S of SOC (state of charge). The method has the advantages that values of the ampere-hour integral processes are repeatedly updated by the aid of the results of extended Kalman filter processes for characteristics of the ampere-hour integral processes and the extended Kalman filter processes under a plurality of conditions of accurate extended Kalman filter processes, and accordingly cumulative errors of the ampere-hour integral processes can be obviously reduced; the results of the ampere-hour integral processes are used as judgment standards, different SOCvalue processes are formulated at different phases of electric quantities, weights of results of two types of algorithms can be automatically adjusted according to the precision of battery models, and accordingly errors can be obviously reduced when the models are inaccurate; as shown by verification, the algorithms are high in accuracy and reliability.