Magnetoencephalography (MEG) is a record of the magnetic fields produced by the electrical activities of the brain using MEG systems. There are three types of sensors for MEG systems: magnetometer and two types of gradiometer. Among them, two types of gradiometer, axial and planar, have been used worldwide. Unfortunately, the waveforms recorded by the two types of gradiometer are often different from each other. This poses a serious problem in comparing and evaluating the data from the two gradiometers. We consider that the MEG study should be published in a way that allows other workers using different types of gradiometer to evaluate and replicate the results of MEG studies. There have been, however, no publication criteria for reports of studies on stimulus-evoked or event-related magnetic fields in human subjects. In this article, we propose publication criteria for evoked or event-related magnetic fields of the human brain: original waveforms of selected channels covering a region of interest, a root mean-squared (RMS) waveform and a contour map at an appropriate time.