A 300 km long magnetotelluric profile across the Great Valley, Sierra Nevada, and California Basin and Range reveals minimum resistivity differences of 1 order of magnitude at depths from Moho at 30–35 km to depths greater than 200 km. Horizontal wavelengths of 25 km are seen for these changes at depths shallower than 100 km, and the Isabella high seismic wave speed anomaly is present in the section between depths of 100 and 200 km as a minimum tenfold resistivity increase. A transformation from electrical resistivity to temperature and melt fraction using a petrological model and laboratory measurements reveals that partial melt is relatively uncommon and confined primarily to depths between 30 and 100 km. The remainder of the observed resistivity variation is attributed to lateral variations in temperature. Predicted melt fractions match those inferred from Miocene-Quaternary xenoliths, but modeled temperatures are ∼200°C higher than inferred from thermobarometric analyses of the xenoliths. Possible causes for this discrepancy include an incorrect starting material for the petrological model, exotic conductive phases, higher conductivity due to oxidation state and/or iron content, and diffusion of ions through the solid matrix. I show that hydrogen diffusion or higher oxygen concentrations are plausible explanation for the higher model temperatures. Less than 200 ppm H/106 Si in olivine would increase the conductivity of the matrix sufficiently to allow temperatures inferred from the resistivity section to match those from the xenoliths. A possible source of the hydrogen would be fragments of sinking, metasomatized mantle lithosphere that melted during Cenozoic delamination.