In order to study the motion of the Pacific plate with respect to the spin axis during the Cenozoic and Upper Cretaceous we combined an analysis of the distribution of pelagic sediments on the sea floor with paleomagnetic data collected on the Pacific plate. The facies distribution of sediments is used to determine the time when an equatorial crossing took place at as many sites as possible. By combining these two independent data sets, bounds can be placed on the relative motion of the 'hot spots' with respect to the spin axis. The data strongly suggest that the Hawaiian hot spot has remained close to its present latitude for the last 70 m.y. with an uncertainty in the rate of latitudinal motion of about 1 cm/yr. A set of continental paleomagnetic data was selected on the basis of the number of samples used, the degree of secondary demagnetization, the uncertainty of the mean direction of magnetization, and the tectonic stability of the region. These data are used to obtain independently the apparent polar wander curve of the Pacific plate. If the plates were rigid and the finite rotations were not grossly in error, reliable continental paleomagnetic poles rotated relative to the Pacific plate would yield a pole path consistent with the sediment facies data and with Pacific paleomagnetic data. Whereas the data from the Pacific plate indicate a latitudinal motion of almost 30¿ during the last 70 m.y., the rotated paleomagnetic poles from the continents cluster around the present-day north pole and require essentially no latitudinal displacement of the plate during this same period of time. Several hypotheses are considered to explain this large discrepancy: a nondipolar behavior of the pleomagnetic field, insufficient quality of the existing Cenozoic paleomagnetic catalogue, or a large amount of internal deformation in the Antarctic or Pacific plate from the Upper Cretaceous to the Lower Oligocene.