We present a thermal model for the lower mantle, which is constructed from petrologically derived estimates of the temperature in the transition zone and from an adiabat based on the thermal properties of MgO and SiO2 measured at high pressures. Superadiabatic contributions to the geotherm through the lower mantle, including those from possible phase transformations but excluding those from thermal boundary layers, are negligibly small. A thermal boundary layer is required at the base of the mantle in order to satisfy our estimate of the lowest possible temperature in the core (about 2800 K); its thickness of about 100 km is well constrained. Even so, our model suggests that we require either a significantly larger heat flux from the core than has been considered reasonable (>50 mW m -2) or the presence of a further thermal boundary layer within the lower mantle in order to arrive at more likely core temperatures (about 3200-3500 K). The latter alternative appears the more plausible, and such a boundary layer requires a barrier to convection. Guided by the present seismological evidence, we suggest that a thermal boundary layer is associated with a chemical discontinuity either at the top of the lower mantle or near its base (D" region). If seismological data require the D" layer to be significantly thicker than about 100 km, this is support for a chemically distinct D" region with multiple thermal boundary layers. Otherwise, the results of this study suggest that the upper and lower mantle may be chemically distinct, hence precluding whole mantle convection.