Due to the importance of battery modeling and characterization and lack of an accurate and comprehensive method, which considers battery as a nonlinear model, this paper introduces a novel methodology for analysis in the frequency domain. This methodology looks to the battery from a different point of view and covers aspects of the battery that is often neglected in the previous work and research studies. Using periodic signals for system identification, allows separating noise and nonlinear distortions from the linear part of the system. Meanwhile random phase multisine signals are very popular as an arbitrary number of frequencies can be added together and applied to the battery at once. In addition to a shorter test time in comparison with conventional single sine EIS (electrochemical impedance spectroscopy), by performing extra periods and different phase realizations, transients are eliminated and noise disturbance and also nonlinear distortion is detected, quantified and qualified. Thanks to the statistical and averaging methods, the linear part of the system can be identified and distinguished from nonlinear noise source, which helps to improve model quality and accuracy. Furthermore this method is used for battery characterization and for evaluating the battery performance and its nonlinear behavior at different current rms values as well as at various state of charge levels. (C) 2016 Elsevier Ltd. All rights reserved.