The use of the NRTL method for correlating experimental data in ternary (liquid + liquid) equilibria is considered. It is concluded that parameters obtained by direct correlation techniques have not a direct physical meaning. Also, it is shown that the resulting values for these parameters depend on the number of experimental points considered and on the particular calculation method employed. Thus, it is very risky to employ such parameters in predicting equilibria of other ternary mixtures. (c) 2004 Elsevier Ltd. All rights reserved.