The invention belongs to the technical field of lithium batteries, and discloses an SOC estimation method and an estimation system for a power battery. The method comprises the steps of establishing apower battery space state model and determining an extended kalman filtering model of the power battery space state model; conducting the test on the dynamic characteristics of the power battery, anddetermining a matching coefficient of the extended kalman filtering model of the power battery space state model according to the test data of the dynamic characteristics; and estimating the SOC according to the matching coefficient of the kalman filtering model. The system depends on the above method. The method and the system can overcome existing technical problems that an existing ampere-hourintegral method requires an accurate initial SOC value and causes the occurrence of error accumulation in the prior art.