This paper presents the results of a new chemical characterization conducted on early Neolithic obsidian artefacts from the excavations carried out in 1971-73 at Pollera Cave, in western Liguria. We re-analysed four artefacts from the impresso-cardial deposit (layer III, level XXII), already analysed by means of neutron activation (INAA) at the end of the ‘70s during Lawrence H. Barfield’s pioneering obsidian circulation research in northern Italy. In addition, three previously unpublished artefacts have been analysed with the same method.The scope of the new characterizations was to solve some problems raised from previous publications, in which contradictory information did not allow us to identify each analysed artefact and attribute it to its original source (at that time determined as Lipari and Sardinia). Therefore, the four artefacts have been re-analysed by non-destructive methods: PIXE at LP2i Bordeaux (AIFIRA platform) and EDXRF at Archéosciences Bordeaux. The results of these analyses were compared to the data obtained with similar analytical methods from obsidian geological samples of potential “island-sources” in the Western Mediterranean. The study established that all the four re-analysed obsidian artefacts from Pollera Cave can now be ascribed to only two different chemical compositional groups (SB2, SC) of the Monte Arci Sardinian source and exclude the presence of Lipari obsidian in the early Neolithic horizons of the cave.These new results outline a picture much more coherent with that emerged from the analyses of obsidian artefacts from the Impressa and Cardial levels of the Arene Candide cave and, more generally, from early Neolithic sites of the Liguro-Provençal arc. Finally, this research pinpoints to the need of systematically reviewing the characterizations published during the first pioneering archaeometric obsidian studies with more precise sourcing databases and methods, to be applied especially when they are in contradiction with a new, more reliable archaeological evidence.< Réduire