Information from river sections in the Hudson Bay lowland indicates that two pre-Holocene nonglacial episodes separated by glacial advances postdate the oldest recognized glaciation. Amino-acid data from in situ and transported marine shell fragments provide relative ages for glacial and nonglacial intervals. Absolute ages for non-glacial sediments as recent as mid-stage 3 were obtained from thermoluminescence (TL) data, although no finite radiocarbon ages have been obtained from wood. Déglaciation and deposition of the Bell Sea marine sediments are correlated to substage 5e by extrapolation from TL data. Ensuing stage 5 glaciation was dominated in Ontario by west-northwestward ice flow emanating from Quebec, and in Manitoba by southwestward ice flow. Deglaciation dated by TL at about 75 ka was followed by isostatic recovery and subaerial exposure in a climate which could have been warmer, but was no more than slightly colder than present. Extensive glaciolacustrine sediments deposited at the close of this interstade were TL dated at about 40 ka in Manitoba. If the TL method has systematically underestimated age, glaciolacustrine sedimentation may date to very late stage 5 or stage 4, or the two nonglacial episodes could be reassigned to substage 5e and stage 7. A resurgence of Quebec-derived ice that culminated as late Wisconsinan glaciation first flowed westward across the entire lowland, but was displaced in the north by southward ice flow. Southwestward and, locally, southward ice flow occurred during final ice retreat along a saddle extending across Hudson Bay and linking domes in Keewatin and Quebec.

Widespread till sheets, glacial lake and glacial-marine sediments on Banks, Victoria, and Melville islands, and on the Beaufort Sea Coastal Plain of the Canadian mainland, may record a late Pleistocene glacial advance which extended to the area as early as the Sangamonian (broad sense) to early Wisconsinan. These sediments overlie beds of interglacial character and underlie in places nonglacial deposits, which have provided both nonfinite and finite ages, and glacial sediments of unquestionable late Wisconsinan age. In other places only a single till sheet is observed between the last interglacial and Holocene sediment suites. Although some workers have argued that the glacial units mentioned above are all late Wisconsinan, stratigraphic, paleoecologic, and chronologic data (14 C, Th/U, and amino acid analyses), from several localities, indicate that the glacial sediments are of likely Sangamonian (broad sense) to early Wisconsinan age and that the nonglacial beds underlying or overlying these date respectively from the Sangamonian and middle Wisconsinan. The dispersal centre during the ice advance was situated, as during other advances in northwestern Canada, west of Hudson Bay. The ice generally extended further during the Sangamonian (broad sense)/early Wisconsinan than the late Wisconsinan but not as far as it did during the early and middle Pleistocene. To help resolve apparent incongruities in interpretation of the late Pleistocene deposits and ice limits it is postulated that extensive Keewatin Sector Ice of the Laurentide Ice Sheet may have first advanced in northwestern Canada during the Sangamonian (broad sense)/early Wisconsinan and remained there until it finally disappeared in the late Wisconsinan.