We examined the early Tertiary global plate velocities with respect to the hot spots and compared these velocities both to those at present and to early Tertiary velocities inferred by assuming that no net torque is exerted on the lithosphere as a whole. In our reconstruction, the velocities of the Pacific and African plates were inferred directly from their hot spot tracks, and the velocities of other plates with respect to the hot spots were calculated from the motion of the Pacific or Africian plate and by relative motions estimated from seafloor spreading data. The relative motion between assumed-separate North Pacific and South Pacific (Chatham Rise) plates was estimated by assuming the hot spots in the Atlantic Ocean basin have been fixed with respect to those in the Pacific Ocean basis, thereby avoiding the use of a poorly defined relative motion circuit. We found that the characteristics of early Tertiary plate motions with respect to the hot spots resemble those at present. The root-mean-square velocity of every major continental plate with respect to the hot spots during the early Tertiary exceeds its present velocity but (with the exception of the Indian plate) is less than the root-mean-square velocity of every early Tertiary oceanic plate. Equatorial lithosphere moved faster than polar lithosphere during the early Tertiary, but the difference is less than at present. Over the interval 64 to 56 Ma, the nonet-torque absolute velocities differ from the assumed fixed-hot spot velocities by only 0.04¿/Ma, which is insignificant. Thus two dissimilar approaches, fixed hot spots and no net torque of the lithosphere, when applied to identical early Tertiary relative plate velocities, yield similar absolute plate velocities. We attribute the differences between these two reference frames found in prior investigations to relative plate velocities different than we assume here.