While temperature drop across the mantle’s basal thermal boundary layer (TBL) is likely $>$1000 K, the temperature anomaly of plumes believed to rise from that TBL is only up to a few hundred Kelvins. Reasons for that discrepancy are still poorly understood and a number of causes have been proposed. Here we use the ASPECT software to model plumes from the lowermost mantle and study their excess temperatures. We use a mantle viscosity that depends on temperature and depth with a strong viscosity increase from below the lithosphere towards the lower mantle, reaching about $10^{23}$ Pas above the basal TBL, consistent with geoid modelling and slow motion of mantle plumes. With a mineral physics-derived pyrolite material model, the difference between a plume adiabat and an ambient mantle adiabat just below the lithosphere is about two thirds of that at the base of the mantle, e.g. 1280 K vs.\ 835 K. 3-D models of isolated plumes become nearly steady-state >10-20 Myr after the plume head has reached the surface, with excess temperature drop compared to an adiabat for material directly from the CMB usually less than 100 K. In the Earth, plumes are likely triggered by slabs and probably rise preferrably above the margins of chemically distinct piles. This could lead to reduced excess temperatures, if plumes are more sheet-like, similar to 2-D models, or temperature at their source depth is less than at the CMB. Excess temperatures are further reduced when averaged over the plume conduit or melting region.