Rates of inter-hot spot motion have been debated for decades. Herein we present updated predictions for the tracks of the Tristan da Cunha, Réunion, and Iceland hot spots assuming them to be fixed relative to Pacific hot spots. Uncertainties in Pacific hot spot rotations, which include uncertainties in the current locations of hot spots of 100-200 km, are combined with uncertainties in relative plate motions accumulated through the plate circuit to obtain the final uncertainty in the predicted positions (including uncertainties of 150-200 km in the current locations of the Indo-Atlantic hot spots). Improvements to reconstruction methods, to relative plate reconstructions, to age dates along the tracks, and to the geomagnetic reversal timescale lead to significant changes from prior results. When compared with the observed tracks, the predicted tracks indicate nominal rates of motion of only 2-6 mm a À1 of these Indo-Atlantic hot spots relative to Pacific hot spots over the past 48 Ma. Within the uncertainties, the rates range from no motion to rates as high as 8-13 mm a
À1. For reconstructions prior to 48 Ma B.P., however, the apparent rates of inter-hot spot motion are much larger, 46-55 ± 20 mm a
À1, if the motion occurred entirely between 68 Ma B.P. and 48 Ma B.P. Either hot spots moved rapidly before 48 Ma B.P., and slowed drastically at ≈ 48 Ma B.P., or global plate circuits through Antarctica become less reliable as one goes increasingly further into the past. Most paleomagnetic data favor the latter explanation.