ABSTRACT.The recent development of large array multichannel biomagnetometers allows the recording of spatiotemporally coherent data sets by simultaneous sampling of magnetic brain activity over a wide extracranial region. The use of such an instrument greatly enhances the speed and the quality of neuromagnetic measurements. To test the reproducibility and validity of neuromagnetic source localization (NMSL), auditory evoked field (AEF) data were repeatedly collected from a single test subject in thirty independent replications. Highly reproducible AEF waveforms, essentially superimposible for the major components M50, M100, and M200, were obtained. When analyzed for source locations using a single equivalent current dipole model, these analyses demonstrated consistent localizations for the sources of M50, M100, and M200 waves with standard deviations of a few millimeters. Clear spatial separation of the sources of the M100 and M200 components was demonstrated. Comparison with MRI showed that equivalent current dipole sources were localized to the region of Heschl's gyrus and the transverse portion of the superior temporal gyrus. Present limitations on NMSL arise from restrictions of the source and body models now in use.