The claim that supercontinents insulate the mantle is largely based on recognition that seismically slow mantle below the Atlantic-African geoid high coincides with the former location of Pangea. We investigate the viability of continental insulation by varying both plate geometries and mantle properties in six three-dimensional (3D) mantle convection models. The efficiency of continental insulation is quantified by calculating the rate of change of the average temperature in the subcontinental mantle. Our findings generally agree with two-dimensional (2D) modeling results. However, we conclude that 2D convection models may exaggerate subcontinental heating, particularly in largely bottom heated cases, and that subcontinental heating results from an absence of subduction in the subcontinental mantle rather than the insulation of active upwellings. Correspondingly, we find that when large degrees of internal heating are present in the mantle, hotter-than-average regions evolve below large oceanic plates. ¿ 1999 American Geophysical Union