Half-space cooling and plate models of varying complexity have been proposed to account forchanges in basement depth and heat flow as a function of lithospheric age in the oceanic realm. Here, werevisit this well-known problem by exploiting a revised and augmented database of 2,028 measurementsof depth to oceanic basement, corrected for sedimentary loading and variable crustal thickness, and 3,597corrected heat flow measurements. Joint inverse modeling of both databases shows that the half-spacecooling model yields a mid-oceanic axial temperature that is>100∘C hotter than permitted by petrologicconstraints. It also fails to produce the observed flattening at old ages. Then, we investigate a suiteof increasingly complex plate models and conclude that the optimal model requires incorporation ofexperimentally determined temperature- and pressure-dependent conductivity, expansivity, and specificheat capacity, as well as a low-conductivity crustal layer. This revised model has a mantle potentialtemperature of1300±50∘C, which honors independent geochemical constraints and has an initial ridgedepth of2.6±0.3km with a plate thickness of135±30km. It predicts that the maximum depth of intraplateearthquakes is bounded by the700∘C isothermal contour, consistent with laboratory creep experiments onolivine aggregates. Estimates of the lithosphere-asthenosphere boundary derived from studies of azimuthalanisotropy coincide with the1175±50∘C isotherm. The model can be used to isolate residual depth andgravity anomalies generated by flexural and sub-plate convective processes.