Measurements have shown a substantial decrease in Northern midlatitude ozone, which has only partially been explained by chemical models. The large ozone depletions determined for the Arctic vortex in recent winters will ultimately spread out and dilute the midlatitudes and thus contribute to the observed decrease. Here we have followed the ozone-depleted air inside the Arctic vortex in 1995 and 1997 during April and May with high-resolution reverse domain-filling (RDF) trajectory calculations. The resulting average midlatitude (30¿--60 ¿N) stratospheric ozone dilution in May is 2.9% and 2.6% of the 1979 column ozone in 1995 and 1997, respectively, or about 40% of the observed depletion. Nearly realistic mixing was introduced by a regridding procedure between successive 7-day long RDF calculations. Low-resolution grid point models give too much mixing, causing an overestimate of the calculated dilution. A recovery of about 12% of the midlatitude dilution in May 1997 is calculated with a photochemical box model, but is not included in the number given above. ¿ 2000 American Geophysical Union