How can we judge whether a non-invasive functional brain image reflects reality? I will talk about using anatomy as a form of ground truth. If we assume that the currents that give rise to the Magnetoencephalographic (MEG) signal derive primarily from the grey matter we can begin to quantify any MEG current estimate. For example: Does the MEG-estimated current distribution follow the cortical surface and if so, at what spatial scale does this relationship break down? Can we find out where the brain is using just the MEG signal? Can we tell the difference between electrical currents originating from different cortical layers? Finally I will talk about the possibility of estimating anatomical structure from MEG data.