If the oldest terrestrial rocks are to be used as a representative sample of the early terrestrial crust, it is important to consider the possibility that they may represent a biased sample. Chemical data indicate that Archean crust is on average depleted in potassium, uranium, and thorium relative to younger crust. Radiogenic isotopes of these elements produce a significant portion of surface heat flow from continents, and low mean heat flow in Archean terrains relative to younger terrains also indicates that these elements are relatively depleted in Archean crust. In stable continental lithosphere the primary variable controlling variations in the geotherm is crustal radiogenic heat production. Qualitative evidence suggests that terrains with low radiogenic heat production, and thus a relatively cool geotherm, are less susceptible to magmatic and tectonic reactivation than terrains with higher heat production and a hotter geotherm. Since crustal radiogenic isotope abundances must have been greater in the past if the isotopes were immobile in stable crust, the relative effect of crustal heat production must have been greater, and thus the range in geotherms must have been greater for the same mantle heat flow. It is suggested that crust with low heat production is less susceptible to orogenic reworking than high heat production crust, and thus there may have been selective stabilization of low heat production (low potassium, uranium, and thorium) crust early in earth history. Either Archean crustal generation was by a mechanism that perhaps fortuitously produced only low heat production crust, or all high heat production crust was reworked by orogenesis during the Archean and did not stabilize in significant volumes until the earliest Proterozoic, perhaps in association with a reduction in mantle temperatures and mean heat flow. If selective crustal reworking did operate during the Archean, surviving Archean terrains may represent a biased sample of the early terrestrial contintental crust.