The advent of large arrays of superconducting magnetic field sensors makes it possible to properly sample the topography of the magnetic field pattern across the human scalp with a temporal resolution of a few milliseconds. These capabilities can be exploited for computing a best estimate of the spatiotemporal evolution of electrical currents within coherently active neural populations distributed across the cerebral cortex. Data from 200 sensors can be interpreted by the method of singular value decomposition to compute a best estimate for the strengths of more than 9,000 current elements that define the magnetic source image moment by moment. Recently, an extension of this technique has been developed to characterize the cortical sources of alpha rhythm. This holds promise for providing a means of identifying the cortical regions that participate in cognitive functions such as mental imagery.