Evoked potentials were recorded to the separate tachistoscopic presentation of a variety of faces and other simple and complex visual stimuli. A positive potential of 150-200 ms peak latency which responds preferentially, but not exclusively, to faces was identified in 8 out of 9 subjects. This potential, best recorded from midline central and parietal electrodes, was evoked by all face stimuli, including photographs, outline drawings, and fragmentary figures. Changes in stimulus size and other parameters which do not affect the clarity of the face, generally had little effect on the peak amplitude. Stimulus changes such as face inversion, reversing the contrast polarity of photographic images, and selectively removing particular facial features, produced a marked increase in latency but often only slight attenuation of this peak. These response properties correspond well with those reported for face-related single cells in the temporal cortex of the rhesus monkey. The scalp distribution of this face-responsive peak also appears consistent with bilateral sources in the temporal cortex.