Olive oils from Libya, Tunisia and Turkey were analysed by gas chromatography using capillary columns (Silar 10C-SiO2-treated surface) with flame-ionisation detection. Data on Italian olive oils were abstracted from the literature. Both sets were stored and converted into a computer-compatible format. The data were statistically treated by FASCLUS and SCATTER methods contained within the G3D procedures of the SAS computer software package. After suitable processing, canonical variables were statistically established from the data sets for the tested oils, leading to the reliable discrimination of virgin and refined oils. It appears that the G3D system is far better than PLOT representation at discrimination. Five unknowns were examined for their authenticity and were classified as expected. The fatty acid methyl ester database was convenient for chemometric techniques and is suitable for sample identification.