The apparent polar wander path for a plate is determined from paleomagnetic data by plotting a time sequence of paleomagnetic poles, each representing the location of the earth's spin axis as seen from the plate. Apparent polar wander paths consist of long, gently curved segments termed tracks linked by short segments with sharp curvature termed cusps. The tracks correspond to time intervals when the direction of plate motion was constant, and the cusps correspond to time intervals when the direction of plate motion was changing. Apparent polar wander tracks, like hot spot tracks, tend to lie along small circles. The center of a circle is called a hot spot Euler pole in the case of hot spot tracks and a paleomagnetic Euler pole in the case of paleomagnetic apparent polar wander paths. Both types of tracks mark the motion of a plate with respect to a point, a rising mantle plume in the case of hot spot tracks and the earth's paleomagnetic axis in the case of apparent polar wander paths. Unlike approaches uced in previous studies, paleomagnetic Euler pole analysis yields all three components of motion—including the east‐west motion—of a plate with respect to the paleomagnetic axis. A new method for analyzing paleomagnetic poles along a track by using a maximum likelihood criterion gives the best fit paleomagnetic Euler pole and an ellipsoid of 95% confidence about the paleomagnetic Euler pole. In analyzing synthetic and real data, we found that the ellipsoids are elongate, the long axes being aligned with a great circle drawn from the paleomagnetic Euler pole to the center of the apparent polar wander track. This elongation is caused by the azimuths of circular tracks being better defined than their radii of curvature. A Jurassic‐Cretaceous paleomagnetic Euler pole for North America was determined from 13 paleomagnetic poles. This track begins with the Wingate and Kayenta formations (about 200 Ma) and ends with the Niobrara Formation (about 87 Ma). Morgan's hot spot Euler pole for 200–90 Ma lies only 15° outside the 95% confidence ellipsoid of the paleomagnetic Euler pole. The good but not perfect agreement reflects displacement between the hot spot and paleomagnetic reference frames at an average rate that is smaller by an order of magnitude than the rate at which the faster plates are moving. The angular velocity of North America about the Jurassic‐Cretaceous paleomagnetic Euler pole was determined by plotting the angular positions of paleomagnetic poles along the track as a function of age. For the Cretaceous the angular velocity was too small to measure. During the Jurassic the angular velocity was high, corresponding to a root‐mean‐square velocity of 70 km/m.y. for the North American plate. A short time interval of even more rapid movement during the Middle and Late Jurassic, possibly corresponding to the beginning of rapid displacement between North America and Africa, is suggested by the data. The direction of absolute motion of North America during the Jurassic was toward the northwest. A Carboniferous‐Permian‐Triassic paleomagnetic Euler pole was determined from 26 paleomagnetic poles. The progression of poles along this track is consistent with known ages and stratigraphy, except for some systematic differences between poles from Triassic rocks on the Colorado Plateau and poles from Triassic rocks off the Colorado Plateau. These differences could be due to a small clockwise rotation of the Colorado Plateau with respect to cratonal North America, or to miscorrelations between Triassic rocks on the Colorado Plateau and off the Colorado Plateau, or to large lag times between the deposition and magnetization of some rock units, or to some combination of these possibilities. Despite these ambiguities in interpreting paleomagnetic data from Triassic rocks, the general pattern of apparent polar wander and plate motion during the Carboniferous through Triassic is clear: The root‐mean‐square velocity of North America was slow (about 20 km/m.y.) during the Carboniferous, probably slow (about 20 km/m.y.) during the Permian, but rapid (60–100 km/m.y.) during the Triassic. Paleomagnetic Euler pole analysis establishes that the present slow (less than 30 km/m.y.) velocity of large continental plates like North America is not an intrinsic property of the plates. Occasionally these plates have, for intervals of 50 ± 20 m.y., moved as rapidly as the oceanic plates are moving today. In our interpretation, during times of rapid motion the continents were attached along a passive margin to oceanic lithosphere that was being subducted at some distance from the continent. Rapid motion stopped when the oceanic lithosphere had been consumed by subduction. If North America, Greenland, and Eurasia were joined as a single land mass during the Jurassic, then a likely location for the subducting oceanic plate attached to this landmass is along the southern margin of the cratonal core of Asia with the oceanic plate extending into Tethys. At the cusp between the Carboniferous‐Permian‐Triassic track and the Jurassic‐Cretaceous track, the trend of the path changes by 160°. The western point of the cusp, which is delineated by paleomagnetic poles from the Chinle, Wingate, and Kayenta formations, is 13° farther west in our analysis than it is in commonly accepted apparent polar wander paths for North America. An implication for terrane analysis is that northward displacements found by using our Late Triassic and Early Jurassic poles are up to 2000 km smaller than are those found by using previously published Late Triassic and Early Jurassic cratonal poles.