The influence of visual fixation position and stimulus size on the scalp-recorded "vertex positive peak" (VPP) evoked by images of faces was studied in three subjects. Responses were recorded, in turn, for line-drawn, frontal-view faces of approximately 8, 4, 2, and 1 deg length, fixated at the centre (bridge of the nose), and at points 1, 2, 3, and 4 deg to the left and right, and above and below, centre. The results showed that central fixation produced VPPs of similar, maximal amplitude for all face sizes. By comparison, "on-face" eccentric viewing yielded attenuated and delayed responses, and the degree of response attenuation as a function of eccentricity was directly related to the face size, with similar amplitude responses being evoked for corresponding fixation locations on each face. Very small or no VPPs were recorded for most "off-face" fixations. Similar results were observed for profile faces, except that the maximal VPP was recorded for fixations near the eyes and not in the centre of the head, and almost identical VPPs were evoked by a centrally fixated face presented with and without an adjacent face or object. These response properties, which correspond to the subjective perception of the facial stimuli, suggest that the VPP reflects brain mechanisms optimized to respond to single, fixated faces, irrespective both of facial image size and of the presence of neighbouring figures.