A number of magnetic variation anomalies observed in geomagnetic deep‐sounding studies can be attributed to inhomogeneities of electrical resistivity in the upper crust. The electromagnetic induction in isolated conductors is usually too small to explain anomalies associated with crustal inhomogeneities, and anomalous variation fields mainly arise from concentration and channeling of currents induced elsewhere. These systems of currents are induced over a large region of the earth and have dimensions comparable to those of the source field. Most of the crustal anomalies are associated with deep conducting sedimentary basins or with regions of rapid change of sedimentary conductivity.Examples of anomalies of this origin are the North German basin, the Anadarko basin in Oklahoma and a region in north‐central Texas where more resistive paleozoic sediments are adjacent to conductive sediments in the Gulf coast plains. Other anomalies are related to currents concentrated in channels of conducting sea water located between two resistive continental blocks. Anomalous fields arising in this way have been observed on both sides of the Bonifacio Straits between Corsica and Sardinia. It is also thought that the Alert anomaly in the Canadian Arctic is closely related to the channeling of current through the Robeson Channel. An additional inland anomaly seems to be due to leakage of currents induced in the oceans into conductive structures in the upper crust. A magnetic variation anomaly with an amplitude larger than any other reported in the literature has been found in the northern Great Plains province of the United States. It extends along the eastern edge of the Black Hills north to the Williston basin. Higher conductivities than those characteristic of sediments are needed to explain this anomaly, which has been tentatively attributed to a zone of conducting graphite schists in the basement. Examples of magnetotelluric observations near current concentrations compared with two‐dimensional model calculations show that plane‐layered solutions are rarely adequate as representations of magnetotelluric resistivity curves in the vicinity of these features. In spite of these difficulties, the geomagnetic deep‐sounding technique can yield information about the lateral variations of resistivities of the upper mantle, providing that the conductivities at the surface are not too high or are relatively uniform