Soil analyses and terrestrial cosmogenic nuclide exposure dating are combined and a conceptual model proposed to explain altitudinal weathering contrasts in high-latitude highlands. We show that summits in the Torngat and Kaumajet mountains were covered by ice during the Last Glacial Maximum, and that their felsenmeer cover probably survived multiple glaciation events. For similar lithologies, soils on felsenmeer covered summits are signigicantly more weathered than those below the felsenmeer limit, displaying higher concentrations of crystalline iron, amorphous aluminium, and silicium extracted with oxalate. Secondary minerals such as gibbsite and kaolinite occur in felsenmeer soils, whereas those formed in till lacked these secondary minerals. 10Be and 26Al exposure ages for nine of ten samples, from high-elevation tors and autochthonous felsenmeer blocks, range from 73 ± 6 to 157 ± 15 ka. By contrast, ages of 11.4 ± 1.0 and 11.7 ± 1.0 ka are measured for bedrock in the much lower Saglek zone, indicating extensive (>3 m) glacial erosion of this zone during Late Wisconsinan glaciation. 26Al/10Be ratios demonstrate that exposure of the high-elevation surfaces was interrupted during at least one cosmic ray shielding event by either ice or till cover. In either case, Late Wisconsinan glaciers could not have extensively eroded these surfaces. Five erratics dated above the Saglek zone, including one in the felsenmeer zone, have exposure ages ranging from 11.6 ± 1.0 to 13.6 ± 0.7 ka. This indicates that valley and high-elevation ice persisted through the Younger Dryas Chron and provides further evidence that the highlands were not nunataks during the Late Wisconsinan period.