Abstract Metasomatic rocks, in which mineralogical change is driven by infiltration of an externally-derived fluid, are challenging for the application of equilibrium thermodynamics because fluid-driven changes in rock bulk composition can be faster than diffusive within-grain re-equilibration. Nevertheless, careful definition of systems in terms of appropriate length scales, an informed choice regarding the controlling variables, and detailed petrological analysis can provide useful results. Thermodynamic calculations using methods that solve for equilibrium between a set of independent end-member reactions produce valid results regardless of the identity of the controlling variables but determination of the relationship between cause and effect in metasomatic rocks is best performed with some knowledge of the parameters that drive the evolution of the system. The correspondence between observations of serpentinised harzburgites from New Caledonia and the results of the application of thermodynamic techniques suggest that equilibrium thermodynamics can be applied to these rocks on appropriate small length scales, varying from micron to cm. The primary drivers for mineralogical change are proposed to be (1) infiltration of a H 2 O fluid, which is buffered to lower a O 2 by the rock with resultant production of H 2 and the resulting change in the redox budget of the rocks; and (2) gradients in a SiO 2 caused by mm to cm scale variation in Si:(Mg + Fe) inherited from the harzburgite protolith.