It is shown how ratios of element mass fractions could possibly be used as markers for origin determination of biodiesel materials of the 1st generation. This is a feasibility study on three samples of soya beans, three samples of canola/rape seeds and the six corresponding final biodiesel products. It is underpinned by a careful validation of the measurement procedures developed. Sample digestion was performed with acids in a high pressure asher or a microwave oven depending on samples. Followed, under ultra-clean laboratory conditions, evaporations to dryness of the digests, re-dissolutions of the solid residues in 2% HNO3 and, for raw plant materials, a final dilution step with 2% HNO3 (1: 100 and 1: 1000). Mass fractions of B, Mg, P, S, V, Cu, Sn, Fe and Zn were measured in all samples by inductively coupled plasma mass spectrometry. These elements were selected from preliminary screening results on 105 and 33 m/z values in biodiesel samples at low and medium mass resolutions, respectively. Selection criteria were: measurement signals better than 10% repeatable and at least 3 times larger than the standard deviation on procedural blanks; and less than 5% relative difference between the measured (not corrected for mass discrimination effects) and natural isotope ratios. Efforts were made at reducing procedural blanks and stabilising their reproducibility to 10-30%. Procedure validation consisted in identifying the factors influencing the results, estimating combined uncertainties and assessing trueness from recovery tests on the certified reference material BCR-191 and simulated biodiesel samples. Recoveries larger than 70% were obtained for [B], [P], [S], [Cu], [Fe] and [Zn]. Ratios of mass fractions were computed for all possible combinations of results obtained for these elements in all samples. These ratios appeared to be good markers of the biological origin of the raw products of this study, with strong correlations between results from plant samples of the same type, and strong anti-correlations between results from plant samples of the other type (R-2 = between 0.86 and 0.96, and from 0.3 to 0.7 when the rapeseed sample was involved). The [S]/[Cu] ratio correlated well too between the biodiesels and their corresponding raw plant samples (R-2 = 0.82 to 0.97). Relative expanded (k = 2) uncertainties on the ratios of mass fractions were estimated to be no larger than 23% in the raw plant materials (and up to 33% for the ratios involving Fe), and 45% for the [S]/[Cu] ratio in biodiesels. (C) 2014 Elsevier Ltd. All rights reserved.