AbstractThermal data represent a valuable remote sensing aid in studying crustal evolution. Heat flow (Q) results from the heat loss by the cooling earth and from the heat production (A) of the radiogenic elements brought to the upper crust by magmatic intrusives. Heat flow is often observed to be linearly related to heat production. The slope of this relation, or thermal depth (D), has been used to infer a global upward enrichment in heat‐producing elements. This thermal depth has been equated with the thickness of granites, but such an interpretation has not been confirmed everywhere. The depth to which granitic plutons are rooted can be computed from the inversion of gravity data. It averages 7±2 km and is much smaller than the thermal depth. Granulite facies rocks are assumed to be present in the lower crust on the basis of seismic and geochemical data. These rocks are generally depleted or initially poor in radiogenic elements (U, Th, K). It is suggested that the thermal depth reflects the depth to the depleted layer in continental regions and that it corresponds to the granulitic layer in most places. Worldwide thermal and seismic data support this relationship, although surface heterogeneities introduce complications. Thermal data can therefore be used to constrain the structure of the crust and its evolution through ti