We have previously argued that the Antarctic and subarctic North Pacific are stratified during ice ages, causing to a large degree the observed low CO₂ levels of ice age atmospheres by sequestering respired CO₂ in the ocean abyss. Here, we suggest a mechanism for the major deglaciations of the late Pleistocene. The mechanism begins with freshwater discharge to the North Atlantic, as evidenced by a Heinrich event, that shuts down North Atlantic overturning. Because of a global requirement for deep ocean ventilation, the North Atlantic shutdown drives overturning in the Antarctic, which, in turn, releases CO₂ to the atmosphere and reduces Antarctic sea ice extent. The resulting increase in atmospheric CO₂ and decrease in albedo then drive global warming and deglaciation. As a control on the timing of deglaciations, we look to the sensitivity of atmospheric freshwater transport to low latitude temperature, which is a natural antagonist to Antarctic stratification under cold climates. While Antarctic stratification is proposed to develop early in a glacial period, continued cooling through the glacial period may reduce the poleward atmospheric freshwater transport and thus may prepare the Antarctic halocline for collapse. Deglaciations may coincide with obliquity maxima because a reduced low-to-high latitude insolation gradient decreases the net poleward freshwater transport and perhaps also because increased polar insolation can warm the deep ocean and shift the westerly winds poleward, all of which should work to weaken Antarctic stratification. Precession minima may encourage Antarctic destratification by biasing tropical water vapor transport toward the northern hemisphere. Finally, obliquity and precession may work together to encourage the circum-North Atlantic freshwater discharge event that initiates the deglacial sequence.