Equilibrium gas compositions for reducing gas mixtures commonly used in the laboratory evaluation of direct reduction (DR) burdens have been calculated using a free energy minimization program (RAND method). The results show that the equilibrium compositions are highly temperature-dependent. In some gas mixtures carbon deposition is possible in the lower range of DR test temperatures normally used. Whereas methane increases the temperature range in which carbon is a favoured species, both water vapour and carbon monoxide supress potential carbon deposition. The thermodynamic results are discussed in connection with laboratory test procedures, and recommendations are given for test gas compositions simulating top-gas and steam reformed natural gas reductants.