Cenozoic global plate motions relative to the hot spots are investigated and compared to plate motions in a mean-lithosphere reference frame. Plate motions were analyzed over six time intervals divided by ages (10, 25, 43, 48, and 56 Ma) chosen, as much as possible, to coincide with key plate reorganizations. Alternative motion circuits and rotational parameters were considered and evaluated with paleomagnetic data from the Pacific and North American plates. The circuit found to be in best agreement with the paleomagnetic data is one in which the hot spots in the Atlantic region are assumed to be fixed relative to the hot spots in the Pacific region. Throughout the Cenozoic, the hot spot and mean-lithosphere reference frames have been in continual, slow relative motion. The rate of motion is nonuniform, however, most of the motion having occurred during the middle Cenozoic. The net Cenozoic rotation of the lithosphere relative to the hot spots is described by a right-handed rotation of 7° about an Euler pole at 46°S, 87°E, which yields a 5° displacement of the north poles of the two reference frames. This motion is small enough that inferences drawn about plate speeds in one reference frame should be valid in the other. Analysis of the global motions resulting from our preferred model showed that many characteristics of current plate motions have persisted throughout the Cenozoic. Plate speeds correlate with latitude, plates moving faster near the equator than near the poles throughout the Cenozoic. As at present, continental plates (except for the Indian plate) moved slower than oceanic plates throughout the Cenozoic. Even the structure of the velocity fields as revealed in a contour of root-mean-square velocities in equatorial bands persists throughout the Cenozoic. The migration of the paleomagnetic axis over time is also compared to the hot spot and mean-lithosphere reference frames. The paleomagnetic axis has shifted 5°–10° relative to the hot spot frame, and a lesser amount relative to the mean-lithosphere frame.