Provenance studies using lead isotopes have two main limitations: the effects of overlaps (some mining regions have signatures that are too close to one another to be clearly discerned) and the difficulty of identifying whether an artifact is derived from a mixture of ores from different sources. These limitations are related to the fact that two-dimensional observation does not allow the full extent of variation in isotope ratios to be observed. This work presents a multivariate statistical treatment: the application of a AHC (Agglomerative Hierarchical Clustering analysis) using the 5 isotope ratios commonly used by archaeologists and geologists. This statistical treatment was tested on two case studies, selected because the origins of the artifacts are well known. It is a question of testing that the statistical treatment creates classes consistent with the contexts of mineralization and links the artifacts to an expected origin (nearby mines in the case of protohistoric copper slags from the Alps, Cypriot rather than Sardinian mining areas in the case of oxhide ingots). In each case, the clustering creates statistical classes corresponding to the mineralized units. The observation of subclasses allows precision at the mining district scale in the case of the Alps but also erases the effects of overlaps between Cyprus and Sardinia. Artefacts that cannot be assigned to a statistical class are potentially derived from a mining zone not included in the database or from mixture of sources. However, in order to confirm this, it seems essential to build a database of the isotopic signatures of the main European copper deposits.