In this paper, a method to estimate the electrochemical impedance under nonlinear conditions using a recursive least squares algorithm is presented. It is based on the estimation, using experimental data, of the Taylor series expansion coefficients of the generalized input-output discrete-time model description of the nonlinear electrochemical system. Using this approach, higher amplitudes of potential variations can be used to identify the frequency response. With such signals, the output signal/noise power relationship is significantly improved leading to lesser measurement time and errors in the results with respect to the classical small input amplitude linear identification. Examples are given in order to illustrate the procedure.