The decline in dissolved oxygen in global oceans (ocean deoxygenation) is a potential consequence of global warming which may have important impacts on ocean biogeochemistry and marine ecosystems. Current climate models do not agree on the trajectory of future deoxygenation on different timescales, in part due to uncertainties in the complex, linked effects of changes in ocean circulation, productivity and organic matter respiration. More (semi-)quantitative reconstructions of oceanic oxygen levels over the Pleistocene glacial cycles may provide a critical test of our mechanistic understanding of the response of oceanic oxygenation to climate change. Even the most promising proxies for bottom water oxygen (BWO) have limitations, which calls for new proxy development and a multi-proxy compilation to evaluate glacial ocean oxygenation. We use Holocene benthic foraminifera to explore I/Ca in Cibicidoides spp. as a BWO proxy. We propose that low I/Ca (e.g., <3 μmol/mol) in conjunction with benthic foraminiferal carbon isotope gradients and/or the surface pore area percentages in foraminiferal tests (e.g., >15%) may provide semi-quantitative estimates of low BWO in past oceans (e.g., <∼50 μmol/kg). We present I/Ca records in five cores and a global compilation of multiproxy data, indicating that bottom waters were generally less-oxygenated during glacial periods, with low O2 waters (<∼50 μmol/kg) occupying some parts of the Atlantic and Pacific Oceans. Water mass ventilation and circulation may have been important in deoxygenation of the glacial deep Pacific and South Atlantic, whereas enhanced remineralization of organic matter may have had a greater impact on reducing the oxygen content of the interior Atlantic Ocean.