The inherent simplicity of cubic EoS models with classic mixing rules allows phase behavior calculations to be performed much faster compared to alternative thermodynamic models. To ensure reliability as well, it is required that temperature-dependent binary interaction coefficients (BIP) are utilized, which however, prohibit the application of the well-known reduction methods. In this work, a new method is proposed for extending the applicability of the reduced variables framework to this type of EoS models. A sampling and interpolation technique is utilized to provide the eigenvalues and eigenvectors at the desired temperature during the simulation run. It is shown that the PPR78 and PR2SRK EoS/g(E) models are perfect candidates for this approach due to the inherent deficiency of their BIP matrix. Nevertheless, the method is directly applicable to any cubic EoS model with temperature dependent BIPs. (C) 2013 Elsevier B.V. All rights reserved.