In paleoceanography, a traditional application of oxygen isotope ratios in fossil carbonate from deep-sea cores concerns the reconstruction of paleotemperatures. Recently, isotopic data have been increasingly employed in the reconstruction of paleosalinities as well. This application involves a number of basic assumptions, a critical review of which is presented here. As the calculation of paleosalinity from δ18O residuals assumes constant and linear surface water salinity (S): δ18O relationships, we investigate this basic assumption for areas with important and variable freshwater budget and/or sea-ice influences. We show that it is particularly unwarranted to assume such temporally invariant linearity in temperate and subpolar seas because of the effects of freezing and melting. More critically, we argue that the determinant of many regions' S:δ18O is advection, not the local water balance, so that the basic δ18O of seawater in many regions of the world ocean would likely have changed in a manner somewhat different from that expected from the ice volume record. Although the differences may not be significant for the gross reconstruction of such features as past sea level, we believe they do potentially bias paleosalinity reconstructions, mostly toward high salinities. To fully exploit the potential of oxygen isotope records as quantitative paleosalinity indicators, both the spatial and temporal variabilities need to be constrained. Hence we recommend that (1) for any time slice of specific interest to paleoceanographic and paleoclimatological studies, a coupled ocean-atmosphere-isotope model, or at least an ocean-isotope model driven by a paleoclimate-isotope model, should be used to define the distribution of the base, mixing end-member, δ18O of sea water; (2) much further regional-scale research is needed on flux rates and isotopic compositions of the various terms in the freshwater budget, as well as on the variability in the various advective terms. The latter variability may be assessed from reconstructions of past seaice distributions and of volume fluxes/isotopic compositions of outflow from marginal seas and through sea straits/passages.