Although the thermal mode of lithosphere cooling can be used to predict the East Pacific Rise topography reasonably well, it does not adequately predict the observed gravity when the zero temperature boundary condition is at the seafloor. The misfit arises primarily because this model predicts thermally lowered densities in the crust at the rise axis which counteract the gravitational effect of the topography, generating a theoretical gravity anomaly smaller than that observed. It is shown, using data at 12¿N as an example, that improved agreement with the thermal model is obtained if, to first order, the crust is treated as an isothermal, constant thickness, constant density layer with variable depth. This result is consistent with seismic data and observations of convective cooling of the crust by seawater circulation. It implies that convective cooling of the whole crust is very rapid, even at the axis of fast spreading ridges. The oceanic Moho may represent the depth to which cracking and convective cooling extends.