Battery state of charge (SOC) is among the most important measures made by an hybrid electric vehicle battery management system since accurate SOC estimation can improve efficiency of power distribution, which can increase fuel economy, extend battery useful life, and ensure balanced pack operation. Existing methods that rely mostly on current integration are prone to sensor bias, causing the resulting estimate to drift away from the true value. This work describes an adaptive model-based method using an equivalent-circuit model augmented with a bias state that is able to correct SOC drift while driving and reduce SOC reset based on open-circuit voltage at key-on.