High power lithium ion batteries are increasingly used in power tools, hybrid electric vehicles and military applications, as a transient power source capable of delivering instant energy, around a relatively fixed state of charge (SOC). Maintaining the battery within pre-specified limits for SOC is important, since lithium ion batteries are prone to safety and/or performance issues during overcharge or rapid discharge below the cut-off voltages. With an increase in the number of cells used in the battery, SOC has a crucial role in cell balancing and optimization of the pack performance. Several techniques have been proposed for the SOC estimation. Most of the existing literature supports an empirical model based on either an electric circuit, arbitrary pole placement or an analytical expression with an arbitrary set of parameters. In spite of their simplicity, the empirical battery models do not provide information on the physical cell limitations. Alternatively, a rigorous electrochemical cell model, aimed at incorporating transport, kinetic and thermodynamic limitations, can be used to estimate parameters that hold a physical significance and hence provide an accurate measure of the cell performance. However, the demand for onboard estimation devices requires estimation techniques that are computationally efficient. In this work, estimation of the SOC of a lithium ion cell using an unscented filtering algorithm is illustrated. The relative advantages and disadvantages of the proposed methodology are briefly discussed. Copyright © 2009 John Wiley & Sons, Ltd.