During the Last Glacial Period (LGP), Greenland experienced approximately thirty abrupt warming phases, known as Dansgaard-Oeschger (D-O) Events, followed by cooling back to baseline glacial conditions. Studies of mean climate change across warming transitions reveal indistinguishable phase-offsets between shifts in temperature, dust, sea salt, accumulation and moisture source, thus preventing a comprehensive understanding of the “anatomy” of D-O cycles (Capron et al,. 2021). One aspect of abrupt change that has not been systematically assessed is how high-frequency, interannual-scale climatic variability surrounding mean temperature changes across D-O transitions. Here, we utilize the EGRIP ice core high-resolution water isotope record, a proxy for temperature and atmospheric circulation, to quantify the amplitude of 7–15 year isotopic variability for D-O events 2–13, the Younger Dryas and the Bølling-Allerød. On average, cold stadial periods consistently exhibit greater variability than warm interstadial periods. Most notably, we often find that reductions in the amplitude of the 7–15 year band led abrupt D-O warmings by hundreds of years. Such a large phase offset between two climate parameters in a Greenland ice core has never been documented for D-O cycles. However, similar centennial lead times have been found in proxies of Norwegian Sea ice cover relative to abrupt Greenland warming (Sadatzki et al., 2020). Using HadCM3, a fully coupled general circulation model, we assess the effects of sea ice on 7–15 year temperature variability at EGRIP. For a range of stadial and interstadial conditions, we find a strong relationship in line with our observations between colder simulated mean temperature and enhanced temperature variability at the EGRIP location. We also find a robust correlation between year-to-year North Atlantic sea-ice fluctuations and the strength of interannual-scale temperature variability at EGRIP. Thus, both paleoclimate proxy evidence and model simulations suggest that sea ice plays a substantial role in high-frequency climate variability prior to D-O warming. This provides a clue about the anatomy of D-O Events and should be the target of future sea-ice model studies.