Although seafloor depth and heat flow for young oceanic lithosphere can be described by modeling the lithosphere as the boundary layer of a cooling halfspace, a long standing question has been why data at older ages deviate from those expected for a halfspace. Two classes of models have been proposed for these deviations. In one, heat added from below ''flattens'' depth and heat flow. In the other, asthenospheric flow beneath the lithosphere perturbs the depth. We compare recent versions of the model classes: the GDH1 thin-lithosphere plate model [Stein and Stein, 1992> and an asthenospheric flow model [Phipps Morgan and Smith, 1992>. The plate model fits heat flow data better than the flow model for all cases considered, and topographic data in all but one case. The flow model significantly overpredicts depths for the North Atlantic, because the assumed asthenospheric flow in the plate motion direction would yield deepening for old ages rather than the observed flattening. Overall, the GDH1 global average model does better than this flow model, whose parameters were fit to specific plates. Moreover, plate models fit to specific plates do better than the flow model. Plate models thus appear more useful than this flow model, suggesting that deviations from a cooling halfspace are largely thermal in origin.