This article discusses the application of thermodynamic and related analysis to reaction systems for enzymic or whole cell catalysis, in which there are high proportions of organic liquid, gas, or supercritical fluid. A variety of predictions may be made, especially based on the partitioning of components between the different phases normally present. In many cases, observed behavior can be explained without invoking any direct molecular effects on the biocatalyst. The predictable changes should always be allowed for before seeking explanations for the residual effects, which are often very different from the crude observations. A sum mary of the general thermodynamics of multiphase systems is presented, and then the main classes of component that distribute between the phases are discussed in turn. Thermodynamic water activity (a(w)) determines the mass action effects of water on hydrolytic equilibria. It also describes the distribution of water between the various phases that can compete in binding water. Because catalytic activity is very sensitive to the hydration of the enzyme molecules, a(w) often predicts an unchanging optimum as other aspects of the system are changed. Hence the a(w) should be measured and/or controlled in these systems, whether the primary aim is to study the effects of water or of other changes. The methods available for measurement and control of a(w) are discussed. Adverse effects of organic solvents or similar nonpolar species partly reflect their tendency to partition into the relatively polar phase around the biocatalyst, especially when this is dilute aqueous. The well-established log P parameter is a measure of this. But other mechanisms of inactivation can occur : directly through contact of the biocatalyst with the phase interface, or indirectly via hydration changes. In these cases the molecular property log P is probably not the best solvent parameter. In low-water systems the biocatalyst remains in a separate phase even when water-miscible solvents are used. Hence, the categorization of solvents in terms of miscibility becomes less relevant. This accounts for the "two peak" dependence of catalytic activity on water content in some miscible systems. Differential solvation of reactants and products, as the bulk phase is altered, causes changes in concentration-based equilibrium constants and yields. These changes in solvation may be monitored through partition coefficient or solubility measurements. Reactant solvation can also account for differences in biocatalyst kinetics, whether or not partitioning into a dilute aqueous phase is involved. These predictable effects should be allowed for when studying effects of solvent or similar changes on activity or specificity. Acidic or basic species (such as reactants) can partition into the microenvironment of the enzyme molecules and adversely affect their protonation state. If a dilute aqueous phase is present, these effects may be analyzed in terms of a pH value, and the problem is simply one of measurement (where the phase is microscopic); some methods are available. At low a(w), it may be more useful to think in terms of direct reaction with protein groups.