In order to ensure that the energy inside the battery is made good use for supplying the portable product and that the risk of damaging the batteries is minimized, the state-of-charge (SOC), which cannot be measured directly, of the battery needs to be accurate. In this thesis, two kinds of estimators, the EKF and the UKF, were introduced and tested in Matlab environment. First, an accurate battery model was quoted and several verified tests were conducted in order to introduce the model at length. The voltage, measured SOC and current data was also obtained in this process for the next estimation step. Then, the EKF and the UKF algorithms were introduced. Experiments were then conducted in order to verify the algorithms in different kinds of situations. The results showed that when the conditions were not ideal, the UKF could perform better than the EKF. Master of Science (Power Engineering)