The N170 is a peak of event-related brain potentials commonly acknowledged to be larger in amplitude for face stimuli compared with any other visual object. Recently, the face selectivity of the N170 has been challenged based on the observation of similar N170 amplitude to faces and cars presented full front. Here, we measured the N170 elicited by the same stimulus categories using a one-back memory and categorization tasks. We found that N170 mean amplitude was significantly larger for cars than faces at electrode sites considered 'optimal' for measuring N170 face selectivity in the absence of task effects. Furthermore, we found evidence for category selectivity and task dependence in the P1 range. These results support the idea that N170 face selectivity is formally questionable.