Aims: To assess the bacterial biodiversity level in bovine raw milk used to produce Fontina, a Protected Designation of Origin cheese manufactured at high-altitude pastures and in valleys of Valle d'Aosta region (North-western Italian Alps) without any starters. To study the relation between microbial composition and pasture altitude, in order to distinguish high-altitude milk against valley and lowland milk. Methods and Results: The microflora from milks sampled at different alpine pasture, valley and lowland farms were fingerprinted by PCR of the 16S-23S intergenic transcribed spacers (ITS-PCR). The resulting band patterns were analysed by generalized multivariate statistical techniques to handle discrete (band presence-absence) and continuous (altitude) information. The fingerprints featured numerous bands and marked variability indicating complex, differentiated bacterial communities. Alpine pasture milks were distinguished from lowland ones by cluster analysis, while this technique less clearly discriminated alpine pasture and valley samples. Generalized principal component analysis and clustering-after-ordination enabled a more effective distinction of alpine pasture, valley and lowland samples. Conclusions: Alpine raw milks for Fontina production contain highly diverse bacterial communities, the composition of which is related to the altitude of the pasture where milk was produced. Significance and Impact of the Study: This research may provide analytical support to the important issue represented by the authentication of the geographical origin of alpine milk productions.