In this study, usual methods commonly applied to report the quantitative composition of flavours, fragrances and essential oils were compared. The MS determinations without response factors exhibited a lack of accuracy and reproducibility. FID provided better results, but the determinations were still unsatisfactory. The best results were obtained by true internal standardization in MS and FID, which is time-consuming for mixtures composed of many constituents. The use of a response factor database was thus proposed and validated by testing the response dependence on GC parameters, using an experimental design. Very low time variations were observed. Under standard GC conditions, the determination of a model mixture with four different instruments exhibited a mean bias of < 3.4%. Copyright © 2008 John Wiley & Sons, Ltd.