The ability of the brain to suppress incoming irrelevant sensory input is termed 'sensory gating,' and auditory sensory gating is often indexed by the auditory evoked response. We recorded the auditory evoked magnetic fields to the human voice, using the conditioning-testing paradigm, to investigate whether or not healthy subjects show less activation to the second voice stimulus. Seventeen healthy adults (mean age 27.9+/-4.8 years, 9 males and 8 females) participated in the experiment. The auditory stimuli were presented monaurally as a series of 120 paired voices, with 500-ms interstimulus intervals and 6-s interpaired stimulus intervals. The P50m and the N100m responses were investigated, and dipole source localization was performed. Root mean squares of both P50m and N100m were significantly suppressed to the second stimulus bilaterally, and the suppression was more significant in N100m. The N100m was located significantly more laterally than the P50m for both hemispheres. These results therefore demonstrate the presence of sensory gating for auditory inputs of the human voice in the primary auditory cortex and the auditory association area.