Multivariate statistical analysis of data characterizing pollution is one of the environmental forensic methods used to search for scientific evidence of the origin of pollution. In 2021, we applied this approach for the canton of Vaud to the analysis of dioxins and furans in soil samples collected in the commune of Lausanne and its surroundings. The levels of 17 congeners – 7 dioxins and 10 furans – analysed in the samples were used to calculate profiles in the form of proportions of congeners constituting signatures of the pollution. The profiles of the samples were then compared to each other by jointly applying a principal component analysis and a hierarchical ascending classification, to determine groups of samples showing similarities or dissimilarities. Explanatory factors for the similarities and dissimilarities were then sought. This treatment of the data made it possible to retain and discard potential sources of pollution suspected by the canton. The conclusions of the statistical study converge with those of the chemical expertise of the data, carried out in parallel to our work. Such an approach can be adapted to other types and contexts of pollution.