The yeast Saccharomyces cerevisiae is one of the most important microorganisms for food and drink production and it is as well a model for biology. Surprisingly, the bases of yeast population structure have been unraveled only recently [1, 2] and the first genomic population approaches have failed to point adaptation to ecological niches. However, as many groups present highly contrasted lifestyle (e.g. anaerobic growth on grape must for wine yeast, aerobic growth as a biofilm on ethanol and glycerol for flor strains, growth on low amount of sugars for oak strains) different adaptations are expected. In this project we have obtained high quality genome sequences of 82 yeast strains: 35 from wine environment (27 wine strains, 8 flor strains) as well as 47 other strains from rum, fermented dairy product, bakery including 8 from oaks trees. Our genomic data enable us to delineate specific genetic groups corresponding to the different ecological niches, and indicate different life cycles. We first detected the amplification of several genes with critical function in their environment (e.g. CUP1 for wine yeast, IMA2 and SUC2 for bakery yeast…). We then established a catalog of genes potentially impacted per population. Several tests revealed a non-neutral evolution at several loci, and especially the presence of selective sweep for wine yeast, suggesting positive selection. The comparison of flor and wine yeast revealed allelic variation associated to growth under velum, as well as the fructophily of flor strains, or a specific metal homeostasis. Last among the three large chromosomal regions originated from horizontal gene transfer (HGT) from distant yeasts first discovered in EC1118 [3] were encountered mainly in wine or flor yeasts but the complete region C was only found among flor strains. These genomic data represent a unique resource for understanding the adaptation of yeast to wine making niches and thus for elucidating the bases of technological properties.