Analysis of interannual variations in polar motion for the period 1963--1991 reveals two prominent components, a quasi biennial component and a lower frequency component of ~4--6 yr period. These frequencies, previously reported in the length of day, are possibly related to the El Ni¿o-Southern Oscillation and Quasi Biennial Oscillation of the coupled ocean-atmosphere system. Over the overlapping period (1980--1991), we compared the geodetically-derived polar motion excitation with equatorial atmospheric angular momentum (AAM). At the biennial frequency polar motion and AAM are well correlated, when the pressure excitation is estimated assuming a non-inverted barometer response of the oceans. However AAM alone is unable to explain the total observed polar motion. An oceanic contribution associated with the dynamical response of the oceans to pressure-derived wind stress is a likely candidate to explain the missing signal. At longer periods (4--6-yr), the east components of polar motion and AAM vary coherently. No clear conclusion can be drawn however for the north components because of large uncertainties reported in the corresponding AAM data at low frequency.