In this work we estimate the state of charge (SOC) of NiMH rechargeable batteries using a robust optimal filter based on a simplified electrochemical model. The robust filter guarantees that the supremum of the error variance - difference between real and estimated SOC - with respect to all admissible uncertainties be minimum. The results are compared with those obtained using the linear Kalman filter. We conclude that both estimations have similar performance, although the robust filter is easier to tune. Experimental results with commercial batteries are provided to illustrate the estimation procedure and it performance. Copyright (c) 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.