Kyser, O'Neil, and Carmichael (1981, 1982) measured theδ 18O values of coexisting minerals from peridotite nodules in alkali basalts and kimberlites, interpreting the nodules as equilibrium assemblages. Based mainly on the systematics revealed inδ 18O-olivinevs. δ 18O-pyroxene diagrams, we have re-interpreted the Kyser et al. data as non-equilibrium phenomena. On suchδ-δ diagrams, the mantle nodules exhibit data arrays that cut across theΔ 18O=zero line; these arrays strongly resemble the non-equilibrium quartz-feldspar and feldspar-pyroxeneδ 18O arrays that we now know arediagnostic of hydrothermally altered plutonic igneous rocks. Thus, the peridotites appear to have been open systems that underwent metasomatic exchange with an external, oxygen-bearing fluid (CO2 magma, H2O, etc.); during this event, the relatively inert pyroxenes exchanged at a much slower rate than did the coexisting olivines and spinels. This accounts for the correlation betweenΔ 18O pyroxene-olivine and the whole-rockδ 18O of the peridotites, which is a major difficulty with the equilibrium interpretation. The metasomatic18O-enrichments of the peridotites can be related to metasomatic enrichments in LIL elements and the development of amphibole and phlogopite. This type of precursor metasomatic activity can explain the development of alkali basalt magmas, as well as leucitites and nephelinites (all of which tend to be slightly18O-rich relative to MORB, withδ 18O=+6 to +7.5). Fluids with appropriateδ 18O values to explain the open-system metasomatic effects can be produced by exchange with ancient subducted oceanic crust (eclogite). However, fluid/rock ratios of about 0.4 to 2.5 are required, indicating that this cannot be a mantle-wide phenomenon. Also, these non-equilibrium effects are apparently transient phenomena, probably associated with the eruptive events that brought the nodules to the surface; at characteristic mantle temperatures, the effects would likely disappear in a few tens of millions of years, or less, implying that the ultramafic nodules are not typical samples of the upper mantle.