Recently, electric vehicles attract great attention for the future mobility. Accordingly, accurate capacity estimation of the batteries becomes more important for safety reasons. To address this, a practical and reliable procedure by model-based approach is introduced for the Li-ion batteries to estimate State-of Charge (SOC) and State-of-Health (SOH) accurately. The idea is that SOH is estimated in charge cycles, while the SOC is estimated by dual extended Kalman filter in discharge cycles with SOH being constant. For further accuracy enhancement, battery model is renewed every time when the SOH is reduced by 10%.