The fluid phase plays an important role in many crustal processes, such as metamorphism, metasomatism, and partial melting, yet little is known about the fluid phase in the deep crust. Early workers assumed that H2O and CO2 were relatively constant with P(H2O) = P(lithostatic) in non-carbonate metamorphic rocks while P(CO2) = P(lithostatic) in carbonates (Turner, 1948). The granulite facies may be an important exception to this traditional view of metamorphic fluids as the lack of hydrous minerals suggests that this may be a metamorphic regime where the fugacity of water is low. Phase equilibria, when applied to fluid buffering reactions, can provide useful information concerning the composition and movement of fluids during the granulite facies metamorphism. The Adirondack Mountains, New York, have been chosen for a study of deep crustal fluids as pressures and temperatures of metamorphism are well known and the area last equilibrated during a single pervasive metamorphism. Two mineral assemblages which buffered the fugacity of H2O (fH2) during the granulite facies metamorphism have been located near the Oregon Dome anorthosite in the Adirondack Mountains, New York. The first assemblage involves the breakdown of amphibole to orthopyroxene, clinopyroxene, quartz and H2O. This assemblage buffered fH2O to low values, with XH2O approximately equal to .1. Another assemblage, which is potentially very useful in the granulite facies, is the reaction of phlogopite and quartz to enstatite, sanidine, and H2O. There are, however, a number of uncertainties which must be evaluated before this assemblage can be applied to calculate fl^O. These include: 1) disagreement between various experiments in the P-T placement of this reaction, 2) a lack of understanding of the relationship of this reaction to more Fe-rich analogues, a problem which is in part due to uncertainties in the determination of the distribution coefficient (KQ) for Mg-Fe between biotite and orthopyroxene, and 3) the possible effects of tetrahedral order or disorder in trioctahedral micas. In spite of these uncertainties this reaction can be applied to calculate the fugacity of water, which is low is these rocks with the average XH2O equal to .1 +/- .1. Two samples contain graphite in addition to an H2O buffering mineral assemblage, making it possible to estimate the fugacity of six fluid species, CO2, H2O, CH4, CO, O2 and H2, if it is assumed that the sum of the partial pressures of these six fluid species is equal to the lithostatic pressure (French, 1966; Ohmoto and Kerrick, 1977). Such calculations indicate that if there was a free fluid phase then-CO2 was the dominant fluid species in these rocks during granulite-facies metamorphism. The other five fluid species were minor. One of the graphite-bearing assemblages is located less then 6 meters from a mineral assemblage which buffered the fugacity of CO2 to low values. This shows that CO2 was not a pervasive fluid in large quantities, but may have been the dominant fluid species in certain rock types. The results of this study indicate that the fugacities of various fluid species can be highly variable and sharp gradients in fluid compositions may exist during granulite-facies metamorphism. Any fluid movement which may have existed in the deep crust would, therefore, be channelized rather than pervasive.