Events surrounding Termination II, as preserved in the Vostok ice core, provide a number of clues about the mechanisms controlling glacial to interglacial climate change. Antarctic temperature and the atmosphere's CO2 content increased together over a period of ~8000 years. This increase is bounded by a drop in dust flux at its onset and by a drop in the Δ18O of trapped air at its finish. A similar lag between dust flux and foraminiferal Δ18O is seen in a Southern Ocean marine record, suggesting that the Δ18O in air trapped in Vostok ice is a valid proxy for ice volume. The synchronous change of atmospheric CO2 and southern hemisphere temperature thus preceded the melting of the northern hemisphere ice sheets. This observation, coupled with the fact that nutrient reorganization in the North Atlantic occurs with or after the sea level rise, eliminates many scenarios proposed to explain the CO2 rise, including those which rely on sea level change, conveyor-related nutrient redistribution, or North Atlantic cooling. Southern Ocean scenarios become the front runners, but the most popular mechanism, iron fertilization, has two problems in explaining the CO2 rise before Termination II. First, much of the dust demise occurs prior to the change in CO2, so if iron is the villain, a threshold value of its supply must be called upon above which productivity does not continue to increase. Second, the CO2 rise continues for some 4--5 kyr after the dust flux has fallen to close to zero. These problems may be solved if the increased iron supply in dust caused higher rates of nitrogen fixation during the glacial periods. In this case the residence time of oceanic nitrate of a few thousand years would enable decreasing productivity to be a global rather than a local phenomenon and would explain the slow rampup of atmospheric CO2. ¿ 1998 American Geophysical Union