In recent industry, hybrid vehicles are gaining more recognition as a practical means for future transportation due to the longer distance, reduced charging time, and less charging stations dependency. The batteries in the hybrid vehicles, however, undergo more complex operation of charge depleting and sustaining modes alternately, which may need more accurate battery state estimation. In this study, a model based method is explored for the Li-ion batteries in the hybrid electric vehicles to estimate State-of Charge (SOC) and State-of-Health (SOH) accurately. While there have been widespread studies for this topic in the batteries research, not many are found that have investigated hybrid operation modes. Also the estimations are mostly limited to normal batteries or shallow degradation with the SOH higher than 90%. In this study, an algorithm based on the dual extended Kalman filter (DEKF) and enhanced self-correcting (ESC) model is developed for the simultaneous estimation of the SOC and SOH. Degradation data for plug-in hybrid vehicle (PHEV) are taken for the study, which undergo the deep degradation of 30%. In order to maintain the accuracy such that the root mean square error (RMSE) of the SOC is within 5% over the entire degradation cycles, two practical methods are proposed: First, the SOH is estimated separately during the battery charging, and is used as a constant in the SOC estimation in the discharging cycles. Second, battery modeling is conducted and the parameters are reset in every intermittent cycles at which the SOH is reduced by 10% initially and by 5% thereafter.