Uplift of old ocean seafloor relative to the familiar t 1 2 curve is generally presumed to be a consequence of an additional heat source at the base of the oceanic lithosphere. Several geodynamical mechanisms have been invoked to explain this behavior. We have taken a petrological rather than geodynamical approach and considered whether the causes of the flattening can be found within the conductive portion of the lithosphere, where viscous creep is insignificant. Accordingly, self-consistent calculations of the phase assemblages of several candidate mantle compositions have been performed, subject to the constraint of experimentally determined calorimetric measurements. We have found that, in general, for temperatures below around 800°C and pressures greater than 10 kbar, the phase change from spinel peridotite to garnet peridotite has a significantly negative Clapeyron slope, which could cause some amount of uplift of old seafloor. The transition, for representative thermal profiles, should be depressed from about 35 km (at 40 Ma) to 45 km (at 160 Ma). Since spinel peridotite is of lower density the net effect is to raise the seafloor topography. The extent of the uplift, which depends primarily on the Al2O3 content of the peridotite and on the effective thermal diffusivity of the lithosphere, should be on the order of 140-220 m.