Neurophysiological data are widely affected by different forms of signal artifacts. In electroencephalographic recordings from the scalp, eye blinks are a main contribution as a source of signal alteration. Different approaches have been used to improve on this problem, from the rejection of part of the signal, to corrections through linear decomposition methods. A widely used technique is independent component analysis (ICA). Different studies have shown the suitability of ICA to correct a variety of artifact sources, but to our knowledge, there is no evidence of the effect of ICA in the phase of a signal, over time. This is of importance because the phase is a critical component of the physiological signals that has been implicated in several neural mechanisms. The aim of this work is to assess the level of phase distortion that ICA can potentially introduce to real and simulated data.