Seasonal and interannual variations of the northern hemisphere stratosphere on the 600 K isentropic surface (20--35 hPa) are investigated by using observations for 1964--1996. An area diagnostic is defined in terms of Ertel's potential vorticity (PV), which measures the area enclosed by PV contours as a function of equivalent latitude and is not tied to the spherical coordinate system. Data from all seasons are examined for oscillations with periods of 6 months to more than a decade. Spectral analysis of PV at equivalent latitudes from 20 ¿N to the pole reveals several possible signals: (1) In addition to the well-known annual cycle of the polar vortex there is a striking semiannual oscillation of PV in middle equivalent latitudes. At 34¿--53 ¿N the semiannual oscillation is larger than the annual cycle. This oscillation arises from the formation of a surf zone in winter, due to planetary wave breaking, superposed on the annual cycle. (2) A signal associated with the quasi-biennial oscillation (QBO) of the equatorial stratosphere, with average period slightly less than 30 months, is strongest at low equivalent latitudes and is apparent in middle equivalent latitudes up to 67 ¿N. Through interaction with the annual cycle, oscillations with periods of 20.2 and 8.5 months are generated. These three spectral peaks at approximately 30, 20, and 8.5 months have been observed in ozone and wind data and are seen here, for the first time, in PV. (3) At middle to high equivalent latitudes, on the other hand, we observe biennial and associated 8-month oscillations related to the occurrence of stratospheric warmings and strong, undisturbed vortices in midwinter to late winter. The data record is remarkable in that, at high equivalent latitudes, strong and weak polar vortices alternated from year to year, producing the biennial signal and a significant negative lag-1-year autocorrelation. The addition of an 8-month harmonic generates a much better fit to the sharp late winter anomalies. (4) South of 60 ¿N a spectral peak at 10.6 years is observed, with the clearest signal near 20 ¿N. This spectral peak may be caused by the solar cycle, or it may arise through interaction of biennial and quasi-biennial oscillations. ¿ 1998 American Geophysical Union