We propose a hypothesis to explain the very abrupt terminations that end most of the glacial episodes. During the last glaciation, the buildup and southerly expansion of large continental ice-sheets in the Northern Hemisphere and extensive cover of sea ice in the N. Pacific and the N. Atlantic imposed a much more zonal climatic circulation system than exists today. We hypothesize that this, in combination with the frigid (dry) polar air led to a significant decrease in freshwater runoff into the Arctic Ocean. In addition the freshwater contribution of the fresher Pacific water was completely eliminated by the emergence of the Bering Strait (sill depth 50 m). As the Arctic freshwater input was depleted, regions of the Arctic Ocean lost surface stability and eventually overturned, bringing warmer deep water to the surface where it melted the overlying sea ice. This upwelled water was quickly cooled and sank as newly formed deep water. For sustained overturn events, such as might have occurred during the peak of very large glacial periods (i.e. the last glacial maximum), the voluminous deep water formed would eventually overflow into the Nordic Seas and North Atlantic necessitating an equally voluminous rate of return flow of warmer surface waters from the North Atlantic thus breaking down the Arctic's zonal isolation, melting the expansive NA sea ice cover and initiating oceanic heating of the atmosphere over the ice-sheets bordering the NA. We suggest that the combined effect of these overturn-induced events in concert with a Milankovitch warming cycle, was sufficient to drive the system to a termination. We elaborate on this proposed sequence of events, using the model for the formation of the Weddell Sea polynya as proposed by Martinson et al. (1981) and various, albeit sparse, data sets from the circum-Arctic region to apply and evaluate this hypothesis to the problem of glacial terminations.