A systematic Lagrangian investigation is performed of wintertime high-resolution stratospheric ozone soundings at Payerne, Switzerland, from January 1970 to March 2001. During the winter season, ozone variability is largest in the lower stratosphere (p > 120 hPa) where the trend is not statistically significant (on the 95%-level) considering the whole time period. A significant negative trend (-5% per decade) exists only above this level. For every ozone sounding, 10-day backward trajectories have been calculated on 16 isentropic levels using NCEP reanalysis data. Both the minimum/maximum latitude and potential vorticity (PV) averaged along the trajectories are used as indicators of the air parcels' origin. A detailed case study of an ozone minihole (198 Dobson units; 1 DU = 0.001 atm cm) in November 2000 reveals that this extreme event was due to the concurrent transport to central Europe of subtropical air in the lowermost stratosphere (where the climatological ozone gradient points to the north) and of polar air in the lower to middle stratosphere (where the ozone gradient is reversed). The importance of transport for the understanding of single ozone profiles is confirmed by the statistical analysis which shows that negative/positive ozone deviations generally coincide with transport from regions with climatologically low/high ozone values. Some important differences arise when considering the layers from 340--440 K and 460--700 K separately. In the lower layer the frequency of transport from the subtropics has increased since 1970 which leads to a (statistically not significant) ozone decrease. The stable relationship between PV and ozone for the 32-year period indicates either no direct chemical impact or no temporal change of this impact. In the upper layer the PV-ozone relationship changes significantly after 1987, and a separate trend analysis for air masses transported from the polar, midlatitude and subtropical regions shows negative ozone trends in all three categories (with a maximum for the polar region). This is not direct evidence for, but would be in agreement with, an increased chemical ozone depletion in the Arctic since the late 1980s. The reasons for the negative trend in the midstratospheric air masses with subtropical origin that are in qualitative agreement with recent satellite observations are presently unknown.