The giant impact model for the origin of the Moon is usually assumed to entail no significant fractionation of nonvolatile lithophile elements relative to a simple binary mixture of impactor silicates plus protoearth silicates. If anything, the most refractory lithophile elements (Al, Ca, Sc, Ti, V, Y, Zr, Nb, REE, Hf, Ta, Th and U) are sometimes assumed to be enriched as a cohesive group. However, the mechanism of giant impact tends to derive the Moon selectively from the peripheral portions of the impactor and the Earth. The Earth may have been hot enough before the impact to be completely molten. particularly if it was insulated by a thick accretionary atmosphere. However, analysis of the likely number and timing of major impacts in the prehistory of the impactor indicates that a fully molten, undifferentiated condition for that relatively small body is unlikely. Given the selective sampling of the giant impact, any significant vertical differentiation within the noncore portion of the impactor would have been largely inherited by the Moon. If the impactor was appreciably bigger than Mars, perovskite probably crystallized along the base of its magma ocean. At lower pressures, majorite would have been important. The process of impactor differentiation inheritance has been modeled, using previous physical models of the giant impact to constrain the depth provenance of the materials contributed from the impactor to the Moon (the protoearth is assumed to be fully molten). Results indicate that if a few tens of a percent of the impactor crystallized prior to the collision, significant fractionations of ratios such as HfYb, ITEHf, ITEZr, CaAl, anTiAl would have been inherited by the Moon (where ITE stands for a variety of incompatible trace elements, most notably Th, U and La). The net fractionation effects are much more pronounced in models in which jetting occurs than in models that do not simulate jetting. On balance, assuming at least a slight amount of jetting occurs, these admittedly imprecise results indicate that the giant impact hypothesis implies a very peculiar composition for the refractory lithophile component of the Moon. The Moon has until now generally been assumed to contain refractory lithophile elements in chondritic proportion to one another; and for some ratios such as Hf/Yb, Sm/Al and Ba/Al, geochemical observations appear to confirm this assumption. However, modeling of impactor differentiation inheritance implies that the giant impact hypothesis and the chondritic-refractory Moon hypothesis are probably mutually incompatible.