For a dipole source, theory predicts 3 useful differences between the MEG and EEG spatial patterns over the head. These are seen when a comparison is made between theoretical MEG and EEG maps, due to the dipole in a spherical model of the head. If true, these differences would allow the MEG to better localize or differentiate neural sources in some ways than does the EEG. A first experimental test of the differences is made here. A comparison is made between MEG and EEG maps due to a neural source which appears to behave as a dipole (N20 of the somatic evoked response). The same 3 differences are seen, therefore the predicted differences are confirmed experimentally. The first 2 differences, due only to the tangential component of the dipole, are that the MEG pattern is rotated by 90 degrees from the EEG pattern and is one-third tighter. The first allows the MEG to localize a tangential dipole better in a preferred direction, across the dipole (while the EEG does so along the dipole); the second allows the MEG to localize somewhat better in its preferred direction than the EEG does in its preferred direction. The third difference is due only to the radial component of the dipole; while the MEG receives no contribution from this component, the EEG pattern is asymmetrical and is heavily weighted by it. This allows the MEG to reveal tangential sources which are obscured by the radial sources in the EEG. For sources which cannot be approximated by a dipole, the MEG-EEG differences will depend on the particular case; however, the spherical model can now be used with more confidence to predict differences in these cases.