The extraction of concentration depth profiles from ARXPS data has been carried out using noisy simulated data and two different approaches, using either simple parametric models or general algorithms with Tikhonov regularization schemes. Among the single parametric models the only one that is stable and robust against noise is that using only one parameter, with the uncertainty of the parameter displaying a linear dependence on the input noise level. For Tikhonov regularization schemes, a guide is given to choose the appropriate regularization parameter, which is based on the use of the S-curve in conjunction with the constraints introduced by chi(2), and which provides user-independent results. (C) 2015 Elsevier B.V. All rights reserved.