Calc-silicate boudins from the Rauer Group, East Antarctica, were metamorphosed under granulite facies conditions during late Proterozoic (ca. 1,000 Ma) m3 metamorphism. Boudin cores contain low to moderate a(CO2) assemblages including wollastonite, grossular-andradite (grandite) garnet, clinopyroxene, scapolite, plagioclase, quartz +/- calcite. Petrological and stable isotopic evidence suggests that these core assemblages resulted from pre-peak m3 infiltration of water-rich fluids; there is no evidence for a pervasive fluid phase under peak m3 conditions. The boudins are separated from the surrounding Fe-rich pelites and semi-pelites by a series of concentric, high-variance reaction zones developed under peak M3 conditions. Variations in mineral assemblage, mineral composition and whole rock composition across these zones suggest that they formed by diffusional mass-transfer, controlled principally by a chemical potential gradient in Ca across the original calc-silicate-paragneiss lithological boundary. As a consequence of the near-complete decarbonation of the calc-silicates before the m3 peak, development of the diffusion-controlled reaction zones did not liberate significant CO2 during granulite facies metamorphism. Similar calcite-poor, low a(CO2) calc-silicate horizons in other granulite facies terrains are unlikely to have been important local fluid sources during deep crustal metamorphism.