Currents induced in Earth by temporal variations in the external magnetic field have long been used to probe mantle electrical conductivity, but almost exclusively from sparsely distributed land observatories. Satellite-borne magnetometers, such as flown on Magsat, Ørsted, and Champ, offer the prospect of improved spatial coverage. The approach we have taken is to isolate induction by harmonic Dst (“disturbance storm time”) excitation of the magnetospheric ring current in satellite magnetic measurements: this is done by removing the magnetic contributions of the main (core) magnetic field, the crustal magnetic field, and ionospheric fields (cause of the daily variation) using Sabaka et al.'s [2000 , 2002] CMP3 comprehensive model. The Dst signal is then clearly evident in the midlatitude satellite passes lower than 50 degrees geomagnetic latitude. At higher latitudes, auroral and field aligned currents contaminate the data. We fit the internal and external components of the Dst signal for each equatorial pass, exploiting the fact that the geometry for the internal and external components is different for the azimuthal and radial vector components. The resulting timeseries of internal and external field variations shows that the Dst signals for the dawn passes are half those of the dusk passes. The sum of equatorial external and internal components of the field averaged over dawn and dusk passes provides an excellent estimate for the Dst index, and may in fact be superior when used as a proxy for the purposes of removing induced and magnetospheric fields from satellite magnetic data. We call this estimate satellite Dst. Cross spectral analysis of the internal and external timeseries shows both greater power and higher coherence in the dusk data. We processed the transfer function between internal and external dusk timeseries to provide globally-averaged, frequency dependent impedances that agree well with independently derived estimates. We estimate Earth's radial electrical conductivity structure from these impedances using standard regularized inversion techniques. A near-surface conductor is required, of thickness less than 10 km with a conductivity-thickness product almost exactly that of an average Earth ocean. Inversions suggest that an increase in conductivity at 440 km depth, predicted by recent laboratory measurements on high pressure phases of olivine, is not favored by the data, although, as in previous studies, the 670 km discontinuity between the upper and lower mantle is associated with a two orders of magnitude jump in conductivity. A new feature in our inversions is a further increase in lower mantle conductivity at a depth of 1300 km. A global map of the internal (induced) component of the magnetic field provides a qualitative estimate of three-dimensional (3-D) variations in Earth electrical conductivity, demonstrating graphically that the satellite data are responsive to lateral variations in electrical conductivity caused by the continents and oceans.