Interannual differences in Arctic ozone are investigated using data from the Polar Ozone and Aerosol Measurement (POAM) III instrument obtained between May 1998 and April 2001. These three winters were unusually warm or cold relative to the 1979–2001 mean, resulting in years with either a warm, disturbed vortex or a cold, quiet vortex. Contours of probability distribution functions (PDFs) of the POAM data are used to identify seasonal and interannual variations in the transport processes controlling ozone. A major warming in December, 1998, displaced the middle stratospheric vortex from the pole for nearly a month, dissipating it. The vortex reformed in January, 1999, filled with high O3 air from lower latitudes, which then cooled and descended. By the end of winter 1999, ozone at 500 K in the vortex was 0.5–1.0 ppm higher than in the subsequent cold winter. The winter of 2000–2001 had frequent wave disturbances, beginning with a fairly large event in November. However, at the end of this event, the high potential vorticity (PV) core of the vortex was intact, and by February, O3 in the vortex looked the same as in 2000. Transport in these winters is inferred from the PDFs and supported by potential vorticity analyses. This study demonstrates that variability in the O3-PV relationship can be caused by transport, independent of loss by polar stratospheric clouds (PSCs). Ozone in the vortex in a very disturbed winter can be unusually high due to transport and will not represent O3 levels expected in the absence of PSCs in other years.