We consider the thermal evolution of the thickened potassium, rare earth element, and phosphorous (KREEP) liquid layer that extends over a significant fraction of the lunar nearside with a prescribed thickness and heat production and beneath an insulating anorthosite crust. This layer represents a subcrustal heat source beneath the Procellarum-Imbrium terrane and may be responsible for the localized and anomalous enrichments of highly incompatible elements in this terrane. In some models this thickened liquid layer is the thermal source for the remelting of mare basalts and also produces the parent magmas of the magnesian suite. Indeed, a KREEP liquid layer thickened from 5 to 10 km with more than 200 times chondritic heat-producing elements does not continue to cool but undergoes a reheating and grows in thickness by dissolving several times its mass of anorthosite and ultramafic cumulates. However, such a liquid layer cannot give rise to the parent magmas to the magnesian suite, because it cannot account for the extraordinarily high incompatible element content, the primitive major element content, or the positive ϵnd of the magnesian suite. This growing layer will also form an impenetrable barrier to the eruption of mare basalts and is generally inconsistent with a number of geophysical constraints of the Moon, specifically, the existence of mascons subsequent to the basin filling by mare basalts. A thickened KREEP liquid layer can, however, explain the asymmetry in the distribution of KREEP-rich rock on the lunar crust. Such a KREEP layer must be efficiently cooled and be quickly solidified to avoid some of the difficulties described above. A thinned anorthosite crust would allow efficient heat loss and induce crystallization of such a layer early in lunar history. ¿ 2001 American Geophysical Union