More than 20 sources of obsidian are scattered across Armenia and the determination of provenance of about 400 artefacts from archaeological sites permits analysis, on a methodological level, of factors which could have influenced the choice of deposit by prehistoric people. The study of the distribution of obsidian shows that there is no simple model: the villages were supplied sometimes from one source, sometimes from several, and in the latter case the nearest deposit was not necessarily the one preferred. The factor of distance as the crow flies, which is often considered a determinant in the choice of sources, is thus often irrelevant. Thanks to a Geographic Information System, we have been able to create a model of 'time-distance' between the deposits and the villages and to establish maps of accessibility to the sources of obsidian from each of the archaeological sites. When several sources are available, a threshold appears which corresponds to the maximum time accepted by the populations for direct procurement; beyond this threshold, the quantity of obsidian is reduced and appears to be related to gradual redistribution or to a transhumance process (revue)