AbstractEach spring, the climatological westerly winds of the Northern Hemisphere (NH)stratospheric polar vortex turn easterly as the stratospheric equator-to-pole temperature gradient relaxes.The timing of this event is dictated both by the annual return of sunlight to the pole and by dynamicalinfluences from the troposphere. Here we consider the predictability of NH final stratospheric warmings inmultimodel hindcasts from the Subseasonal to Seasonal project database. We evaluate how well theSubseasonal to Seasonal prediction systems perform in capturing the timing of final warmings. Wecompare the predictability of early warmings (which are more strongly driven by wave forcing) and latewarmings and find that late warmings are more skillfully predicted at longer lead times. Finally, we findsignificantly increased predictive skill of NH near-surface temperature anomalies at Weeks 3–4 lead timesfollowing only early final warmings.