Questions concerning the initiation and growth of the continental ice sheets at the onset of the last glaciation are explored by comparing Milankovitch forcing, paleoclimatic records, and the results of global circulation model experiments. Decreases in the solar radiation at lat 50° and 60°N over the past 140 ka provide an apparent reason for the cooling intervals, which, in accord with paleoclimatic evidence, started with a dramatic and rapid temperature decrease ca. 115 ka. Subsequent insolation decreases, although in phase with paleoclimatic evidence of cooling, appear to be unrelated to the magnitude of the response, particularly during marine oxygen-isotope stages 4 and 2. The gradual and relatively small insolation changes during stage 3 could not have alone caused the rapid climatic oscillations evident in ice cores, pollen records, and deep-sea sediments. The magnitude of the Holocene wanning appears to be greater than that generated by an equivalent insolation regime 80 ka.

A series of experiments performed using the Goddard Institute for Space Studies (GISS) general circulation model driven by modified insolation input corresponding to the time of the initiation of the last glaciation failed to maintain snow cover at high latitudes during the summer, despite reduced summer and fall insolation at 116 ka. Even when the differences between the insolation fields at 116 and 114 ka were magnified by a factor of five, snow did not accumulate. When 10-m-thick ice was added in the model to locations where continental ice sheets existed during the last glacial maximum, the model failed to maintain this ice; melting rates over North America were such that the ice would have disappeared within five years. Only when the sea-surface temperatures and carbon-dioxide concentrations were reduced to the full glacial values was the model capable of maintaining the ice, but only in a restricted region of northern Baffin Island.

Modeling results and evaluation of the paleoclimatic records imply that many key questions concerning the response of the atmosphere-ocean-biosphere-cryosphere to orbital forcing remain unresolved. Mechanisms and feedbacks involved in the general pattern and details of global climate change can be defined only as chronological resolution of the paleoclimatic record is improved.