A series of numerical experiments on mantle flow and melting predict a positive relation between mantle temperature and crustal thickness. The models also demonstrate that crust formed at slow spreading rates is more sensitive to variations in mantle temperature than crust formed at fast rates so that the range of calculated thicknesses is much greater for crust formed at slower rates. An instantaneous mantle temperature increase results in a transient pulse of melt production that is also more pronounced at slower spreading rates. The predicted behavior is caused by the interplay between mantle flow driven by plate separation and that driven by thermal, compositional, and melt-related buoyancy. A temperature increase results in a decrease in mantle viscosity and an increase in the depth at which melting begins. A lower viscosity leads to stronger buoyancy-driven flow that carries more mantle to shallow depths below the ridge. Thermal buoyancy effects, which may result in cooling and mixing of depleted and undepleted material under the ridge, appear to be of greater importance at slower spreading rates. The steady state results are broadly consistent with global compilations of oceanic crustal thickness that show larger variations in crustal thickness at slower spreading rates than at faster rates. thicknesses estimated from seismic refraction data from crust formed within a single segment of the Mid-Atlantic Ridge but at different spreading rates (1.0 to 1.9 cm/yr) rate consistent with (but do not prove) the model results. The transient pulse of melt production associated with a rapid increase in mantle temperature might occur when a ridge becomes proximal to a hot spot.