The ability of the DISQUAC model for predicting excess enthalpies (HE) of ternary systems on the basis of binary parameters only, i.e., neglecting ternary interactions is analyzed. At this end, DISQUAC results for a set of 95 ternary systems are examined. The solutions studied are formed by only hydrocarbons (or CCl4); or by one polar compound (not alcohols) and two hydrocarbons (or CCl4); or by two polar compounds (not alcohols) and one hydrocarbon (or CCl4); or by one alcohol and two hydrocarbons (or CCl4). Most of the HEs analyzed are endothermic, and valid at 298.15 K and atmospheric pressure. The mean deviation between experimental values and DISQUAC results is 5.5% for the ternary systems and 6.5% for the constituent binaries (181 mixtures). The interaction parameters used are valid for the description of thermodynamic properties of binary systems: vaporliquid equilibria (VLE), liquidliquid (LLE), and solidliquid equilibria (SLE), HE and excess heat capacities at constant pressure (CEP), as well as HE and VLE of ternary solutions. Predictions are, in most of the cases, independent of the mixture compounds, or the number of groups present in the system. Larger deviations underline typical shortcomings of the group contributions methods (e.g., Patterson's effect; branching). Note that results for the ternaries and for the constituent binaries are of the same order. This is not the case for the Dortmund version of UNIFAC. The mean deviations obtained using this model are 10.5% and 14% for the ternary and binary mixtures, respectively. Results from other models (original UNIFAC, Flory's theory, NittaChao, UNIQUAC association model) for a number of systems are also compared to those obtained using DISQUAC.Key words: predictions, HE, ternary systems, binary parameters, geometrical methods, Flory, group contributions.