In oceanic regions the effective elastic thickness Te of the lithosphere is generally estimated from the free air gravity field and approximately corresponds to the depth of the 450 ¿C isotherm. In continental regions the Bouguer anomaly is commonly used and gives thicknesses for shields as great as 130 km, where geotherms reach temperatures of 800--1000 ¿C. If this result were correct, it would require the continental lithosphere to be able to support elastic stresses at considerably higher temperatures than the oceanic lithosphere can. However, detailed examination of the free air and Bouguer anomalies over continents shows that values of Te obtained from Bouguer anomalies over shields are often upper bounds, rather than estimates, because most of the short-wavelength topography has been removed by erosion, and what remains is incoherent with both the free air and Bouguer gravity anomalies. The present study uses free air anomalies, and topography as a load whose geometry is known. The main importance of subsurface density contrasts and their associated gravity anomalies is that they reduce the signal to noise ratio, without affecting the estimated value of Te. No estimates of Te from free air gravity and topographic data from North America, East Africa, Australia, the Indian peninsula, and the former Soviet Union exceed 25 km, and all are smaller than the seismogenic thickness. The largest estimate, of 42 km, comes from profiles across the Himalayan foredeep. The gravity in this region may be affected by dynamic effects. There is therefore no evidence that the continental lithosphere can support elastic stresses for geological times at temperatures above about 350 ¿C. ¿ 1997 American Geophysical Union