This paper is concerned with an analysis of the effect of the Pleistocene glacial cycle upon the Earth's rotation. We demonstrate that two important geophysical observables may be explained as aspects of the rotational response to surface mass loading by ice sheets. These are the astronomically observed non-tidal component of the acceleration of planetary rotation and the secular drift of the rotation pole relative to the surface geography which is evident in the ILS pole path. The former observation is shown to provide an unambiguous constraint upon the viscosity of the planetary mantle and requires a depth dependence of this parameter which is the same as that which has previously been inferred in studies of the relative sea-level variations and free air gravity anomalies associated with post-glacial rebound. The observed secular drift of the pole cannot provide an independent estimate of the viscosity of the mantle because it depends jointly upon mantle viscosity and lithospheric thickness. With the viscosity profile fixed by the rebound data, the observed speed of polar wander suggests a continental lithospheric thickness which is in excess of that appropriate for old ocean basins but nevertheless in accord with independent constraints.