Two effects that contribute to the subduction of continental lithosphere are the negative buoyancy of the relatively cold mantle part of continental lithosphere and the pull of a downgoing slab of oceanic lithosphere on continental lithosphere trailing behind it. The mantle part of the lithosphere could continuously subduct about 10 km of continental crust, if the upper and lower crust could be detached from one another. This estimate, however, could be in error by as much as a factor of three, given the range of plausible values for the requisite parameters. The second effect is more difficult to estimate because of our ignorance of what proportion of the gravitational force acting on the downgoing slab is transmitted to the surface lithosphere. For various assumptions, subducted oceanic lithosphere could be expected to pull a short (tens of kilometres) or long (hundreds of kilometres) length of intact continental crust into the asthenosphere. Although we favor subduction of only a short length of intact continental crust, we cannot prove that a large amount of crust is not subducted. These calculations are for continental margins that meet subduction zones flush. If continental crust can remain intact, peninsulas and microcontinents might be subducted completely.