The reconstructed positions of many of the Indian Ocean intraplate volcanic features are inconsistent with the present positions of the hotspots to which they have been ascribed. The discrepancy can be explained by the past motion of the hotspots themselves, and in the errors introduced in an absolute plate model based on fixed hotspots. Previous models for the motion of hotspots in the Indian Ocean (e.g. [Steinberger and O'Connell, in: The History and Dynamics of Global Plate Motions, 2000, pp. 377-398, and Geophys. J. Int. 132 (1998) 412-434; Antretter et al., Earth Planet. Sci. Lett. 203 (2002) 635-650]) have had success in matching paleolatitudes of the Kerguelen hotspot, but not the Marion or Reunion hotspots. We calculate the motion of the Indian Ocean hotspots using a method described in Steinberger and O'Connell (op. cit.), employing a plate motion model revised using identification of Mesozoic magnetic anomalies in the Enderby Basin [Gaina et al., in: The Evolution and Dynamics of the Australian Plate, in press]. We find a motion of ∼7° south for the Kerguelen plume, similar to previous models [Antretter et al., op. cit.]. This motion alleviates the discrepancies in the position of the Kerguelen plume and its purported involvement in the formation of the Ninetyeast Ridge, Rajmahal Traps, Bunbury basalts and assorted volcanics of the Western Australian margin. We find a motion of 5° north and 7° south for the Reunion and Marion plumes, respectively. The motion of the Reunion hotspot is consistent with paleolatitude estimates. The apparent fixity of the Marion hotspot [Torsvik et al., Earth Planet. Sci. Lett. 164 (1998) 221-232] is not well-constrained, but may be due to the combination of hotspot motion and true polar wander acting in opposing directions. We present revised absolute finite rotations for the Indian plate for the last 65 Ma based on the motion of these hotspots, which place India farther north in the past than for fixed hotspot models.