The present study compares the ability of two semi-predictive approaches, namely the global phase diagram approach (GPDA) and the predictive Soave-Redlich-Kwong model (PSRK), for describing the experimental data in the binary homologous series of n-alkanes. A method to expand the application of the GPDA model to the heavy n-alkanes, absent in the Design Institute for Physical Property Data (DIPPR) data base, is proposed. Since both models do not implement the binary data of the systems under consideration for evaluation of their parameters, they appear here as entirely predictive. It is shown that both models are reliable in predicting the data of symmetric systems and they yield deviations that do not significantly exceed the possible experimental uncertainties. The robustness and reliability of GPDA in comparison with PSRK becomes evident predicting the data of asymmetric systems. PSRK tends to overestimate the liquid-liquid immiscibility range and as a result it over predicts the phase equilibria pressures and fails to describe the global phase behavior. In contrast, GPDA describes the global phase behavior exactly and yields accurate predictions of both the critical and the sub-critical data even for very asymmetric systems, such as propane-n-hexatetracontane. (C) 2003 Elsevier Ltd. All rights reserved.