The average crustal thickness of the southern highlands of Mars was investigated by calculating geoid-to-topography ratios (GTRs) and interpreting these in terms of an Airy compensation model appropriate for a spherical planet. We show that (1) if GTRs were interpreted in terms of a Cartesian model, the recovered crustal thickness would be underestimated by a few tens of kilometers, and (2) the global geoid and topography signals associated with the loading and flexure of the Tharsis province must be removed before undertaking such a spatial analysis. Assuming a conservative range of crustal densities (2700--3100 kg m-3), we constrain the average thickness of the Martian crust to lie between 33 and 81 km (or 57 ¿ 24 km). When combined with complementary estimates based on crustal thickness modeling, gravity/topography admittance modeling, viscous relaxation considerations, and geochemical mass balance modeling, we find that a crustal thickness between 38 and 62 km (or 50 ¿ 12 km) is consistent with all studies. Isotopic investigations based on Hf-W and Sm-Nd systematics suggest that Mars underwent a major silicate differentiation event early in its evolution (within the first ~30 Ma) that gave rise to an enriched crust that has since remained isotopically isolated from the depleted mantle. In comparing estimates of the thickness of this primordial crust with those obtained in this study, we find that at least one third of the Martian crust has an origin dating from the time of accretion and primary differentiation. Subsequent partial melting of the depleted mantle would have given rise to the remaining portion of the crust. While we predict that a large portion of the crust should be composed of ancient enriched materials, a representative sample of this primordial crust does not currently exist among the known Martian meteorites.