The impact of the final drainage of Lake Agassiz, some 8.4 ka ago on the Atlantic Meridional Overturning Circulation is still debated. The lack of isotopic response in planktic foraminifer records at the outlet of the drainage channel constitutes a puzzling element in this debate. To estimate the response of the 18O–salinity relationship in ocean surface waters to drainage, we use a core raised from southeast Hudson Bay to document δ18O values of lake waters, immediately prior to drainage. Valves of Candona sp. (an ostracod) from lacustrine varved-sediments underlying marine clays, yielded δ18O values clustered around –20.5‰ (versus VPDB (Vienna PeeDee Belemnite)). Assuming a lake bottom temperature of about 0 °C, an isotopic composition of –25‰ (versus VSMOW (Vienna standard mean ocean water)) is calculated for paleolake waters, not unlike values reported for earlier stages of Lake Agassiz. This value is only slightly more negative than those estimated for Laurentide Ice Sheet (LIS) meltwaters during the last glacial maximum (∼ –21‰), or for the “apparent” freshwater end member diluting modern northwest Atlantic surface waters (–20.3‰ ± 0.4‰). The estimated ∼ –0.1‰ shift of the 18O–salinity relationship of ocean surface waters, in the salinity domain (>34) characterizing the only planktic foraminifer (Neogloboquadrina pachyderma, Np) then abundant off the drainage outlet, seems barely distinguishable from background noise of Np δ18O records from such environments. Most other similar LIS deglacial events have probably not left any clear isotopic records in deep-sea cores from surrounding basins.