Petroleum fractions are typical complex mixtures. For the simulation of petroleum refining processes they are characterized by substitute mixtures. In this contribution the traditional approach based on pseudocomponents is compared with the newer approach utilizing a Substitute Mixtures of Real Components (SMRCs). The main advantage of SMRC approach is the direct availability of physical property data and the knowledge of the chemical character. The most important task in many applications is to characterize the C7+ fraction. It is shown, how it can be characterized by a small number of real components. The examples used for comparison comprise separation processes and also the pyrolysis modeling, where the known chemical character of a SMRC is crucial for feeding an artificial neural network model. (C) 2012 Elsevier Ltd. All rights reserved.