The motion of hotspots and the deformation of their underlying plume conduits as calculated within models of global mantle flow are presented. A new list of 44 possible hotspots with associated tracks has been compiled. For all of them, calculations have been performed under consideration of individual age and anomalous mass flux for three different models of plume buoyancy and mantle flow. Plume source depth has usually been assumed to be the top of D″, but an alternative source depth at the 670-km discontinuity has also been considered. Using models of relative plate motions and boundaries, hotspot tracks on plates have been calculated and compared with age data, ocean floor topography, and distribution of volcanics on continents. Absolute plate motions have been redetermined under consideration of hotspot motion, using a new least squares method. For the Hawaiian and Yellowstone hotspots, source locations and hotspot motion have been computed for a total of up to 23 different models. The results show plume conduits being tilted, with source regions at the D″ moving in the lowermost mantle flow, generally toward large-scale upwellings under southern Africa and the south central Pacific. Hotspot surface motion often represents the horizontal component of midmantle flow, which is frequently opposite to plate motion, toward ridges and away from subduction zones. In particular, almost all models tested predict southward motion of the Hawaii and Kerguelen hotspots and westward motion of the Iceland hotspot. For models including hotspot motion the agreement between calculated and observed hotspot tracks is frequently about as good as, or better than, for the fixed hotspot model, but sometimes fixed hotspots give the best fit. In some cases where the track ends at a subduction zone, e.g., for the Bowie hotspot, results can give indications about the otherwise unknown age of the hotspot. In other cases, especially for the Tahiti hotspot, results suggest an origin shallower than D″, and in yet other cases, particularly East Africa, the failure of the hotspot models used supports other evidence indicating the presence of comparatively broad upwellings rather than localized plumes. ¿ 2000 American Geophysical Union