A state of charge (SOC) algorithm is described and implemented for a nickel metal hydride (NiMH) battery system. The essential elements of the algorithm were first used as part of the GM Precept hybrid electric vehicle program, which culminated in an 80-mile-per-gallon, 5-passenger, technology-demonstration vehicle. The algorithm is based on a simple equivalent-circuit representation of the battery pack, with the parameters adapted by means of on-line least-square regression. The output of the algorithm is the SOC, which is a measure of the available energy within the battery system, and the power capability for subsequent charge or discharge, which enables efficient hybrid vehicle operation through optimal torque allocation. The behavior of the algorithm in terms of convergence, accuracy, and robustness is examined; for these analyses, the power required from the battery for the Precept to complete a variety of drive schedules provides the power versus time trace. Voltage hysteresis is shown to play a critical role, and this effect makes determining the SOC particularly difficult for NiMH batteries relative to other battery systems (e.g., lead acid and lithium ion). (C) 2003 Elsevier B.V. All rights reserved.