A combined geophysical-petrological methodology to study the thermal, compositional, density, and seismological structure of lithospheric/ sublithospheric domains is presented. A new finite-element code (LitMod) is used to produce 2-D forward models from the surface to the 410-km discontinuity. The code combines data from petrology, mineral physics, and geophysical observables within a self-consistent framework. The final result is a lithospheric/ sublithospheric model that simultaneously fits all geophysical observables and consequently reduces the uncertainties associated with the modeling of these observables alone or in pairs, as is commonly done. The method is illustrated by applying it to both oceanic and continental domains. We show that anelastic attenuation and uncertainties in seismic data make it unfeasible to identify compositional variations in the lithospheric mantle from seismic studies only. In the case of oceanic lithosphere, plates with thermal thicknesses of 105 ± 5 km satisfy geophysical and petrological constraints. We find that Vp are more sensitive to phase transitions than Vs, particularly in the case of the spinel-garnet transition. A low-velocity zone with absolute velocities and gradients comparable to those observed below ocean basins is an invariable output of our oceanic models, even when no melt effects are included. In the case of the Archean subcontinental lithospheric mantle, we show that "typical" depleted compositions (and their spatial distribution) previously thought to be representative of these mantle sections are compatible neither with geophysical nor with petrological data. A cratonic keel model consisting of (1) strongly depleted material (i.e., dunitic/harzburgitic) in the first 100-160 km depth and (2) less depleted (approximately isopycnic) lower section extending down to 220-300 km depth is necessary to satisfy elevation, geoid, SHF, seismic velocities, and petrological constraints. This highly depleted (viscous) upper layer, and its chemical isolation, may play a key role in the longevity and stability of cratons. copyright 2008 by the American Geophysical Union.