The separation units of the biodiesel production process require thermodynamic models able to correlate and predict the phase equilibria of mixtures at different stages of the process. Moreover, some units like those in the glycerol washing sector, require high accuracy due to their sensitivity to the components distribution coefficients. In this work, the Group Contribution Equation of State (GCA-EoS) is applied to model several systems of importance in this field. The mixtures evaluated can be highly non-ideal, especially in the last purification steps where polar compounds are infinitely diluted in biodiesel. Experimental VLE and LLE data were satisfactorily correlated and predicted with a single set of interaction parameters, using a simplified version of the association contribution of the model. In this way GCA-EoS can be used for the whole process with a faster convergence of simulated units in comparison with the case of considering the rigorous association model. Furthermore, taking into acount that GCA-EoS is a group contribution model, biodiesel production from different raw material, i.e. soy, palm or canola oil, can be predicted without requiring new parameters adjustment. The prediction model can be applied to the design, simulation and optimization of the production and purification sections. (C) 2013 Elsevier B.V. All rights reserved.