An irregular distribution of mass on the surface of the earth can cause a reorientation of the rotation axis with respect to the earth, i.e., true polar wander. To cause an instability, there must be regions with vertical density contrasts such that there are different contributions per unit area to the inertia tensor of the earth, and also the geometric arrangement of these regions must be such that the off-diagonal components of the inertia tensor are nonzero. In general, isostatic compensation results in first-order cancellation of inertia effects, and the smaller second-order effects will be determined by the specific density distribution. It has been suggested that the re-arrangement of the continents and oceans has caused polar wander, but calculations of inertia contributions for density models give near cancellation. The ridges and trenches may have greater potential for causing imbalances. The cold dense slab in the trenches creates a mass excess and thus an inertia excess over normal oceanic lithosphere. A thermal model of an isostatic ridge also gives a positive effect. Excitation tensors are constructed for the ridges and trenches for the present and for three times since the Early Cretaceous to determine the position of the principal axes. Neither the system of ridges nor trenches alone give principal axes near those of the earth for the the present. The combination of the ridges and trenches with principal axes nearest the earth's has the ridges weighted equally in magnitude but opposite in sign to the trenches. This weighting is discordant with density models and the sign of gravity and geoid anomalies. A more reasonable weighting, equal and same sign, gives principal axes with no relation to those of the earth.