In this paper, we have tried to classify 382 samples of gasoline and gasoline fractions by source (refinery or process) and type. Three sets of near infrared (NIR) spectra (450, 415, and 345 spectra) were used for classification of gasolines into 3 or 6 classes. We have compared the abilities of three different classification methods: linear discriminant analysis (LDA), soft independent modeling of class analogy (SIMCA), and multilayer perceptron (MLP) - to build effective and robust classification model. In all cases NIR spectroscopy was found to be effective for gasoline classification purposes. MLP technique was found to be the most effective method of classification model building. (C) 2007 Elsevier Ltd. All rights reserved.