The use of H-1 NMR in combination with principal component analysis (PCA) is proposed, for the first time, to discriminate methyl biodiesel from different sources. Sixty-six samples, either pure or combined into binary blends of biodiesel from castor, cotton, peanut, pinion, soybean and tallow were used in this study. The PCA of the selected regions in H-1 NMR spectra proved to be a powerful tool for determining biodiesel present in blends, and it was also very effective in identifying their sources. Furthermore, by selecting appropriate regions in H-1 NMR spectra, it was possible to predict the biodiesel with different fatty acid methyl esters (FAME) compositions and contents in blends. Moreover, it was possible to correlate the biodiesel composition with properties such as viscosity and crystallisation temperature of the FAME blends. However, due to similarities in the fatty acid content and the absence of notable spectral differences in H-1 NMR, it was not possible to distinguish pinion and peanut biodiesel blends. These blends were separated of the analysed group by elimination criteria, and their pure biodiesel (B100) was identified by the characteristic diallylic hydrogen chemical shifts for the linoleic esters. (C) 2012 Elsevier Ltd. All rights reserved.