This study investigated polycyclic aromatic hydrocarbons (PAHs) and hydrocarbon oil index (HOI) pollution in the soil on the territories of two former primitive asphalt pavement plants (APPs) in Estonia. The standard quantitative methods for the chemical characterisation of the oils consisted of an initial screening, by using a gas chromatography-flame ionization detector (GC-FID), and, for a more detailed analysis, of gas chromatography-mass spectrometry (GC-MS). A combination of chemometric and analytical methods was used to identify the sources of PAHs, which could be attributed to the soil pollution at the plants. The identification and classification of oil spills were performed using chemometric techniques, such as the principal component analysis (PCA) and the clustering analysis (CA), which is based on Jaccard similarity. The application of the chemometric techniques enabled the clustering and discrimination of polluted soils into four groups, according to oil type. Several different methods of CA, such as single, complete and average linkages, were tested and the results were compared.