Aims: To analyse the diversity of wild yeast in spontaneous fermentations of a white wine and to select the most suitable autochthonous starter yeasts. The selected yeasts would be used for inoculation of industrial fermentations in several years. Methods and Results: Yeasts were characterized by applying electrophoretic karyotyping. This technique was chosen because it can reveal the large-scale mutations in the yeast genome induced by gross chromosomal rearrangements. This type of mutation is considered one of the main forces behind the rapid evolution of industrial yeasts. A heterogeneous population of yeast strains was observed in the spontaneous fermentations during two consecutive years. Four of the most abundant strains were isolated and tested for microbiological features of industrial importance. The selected autochthonous strains were used as starter yeasts for the following 7 years. In the majority of these experiences, we obtained homogeneous yeast populations, in which the karyotype of one of the inoculated strains - karyotype V - emerged as clearly dominant. Conclusions: The inoculation of the selected strain with karyotype V and a proper handling of the inoculum scaling-up process led to the substitution of the spontaneous fermentations by controlled fermentations producing a highly satisfactory final product. Significance and Impact of the Study: We monitored the wine yeast population of an industrial system for a total of 9 years. Our work is one of the first examples made at industrial scale showing how molecular techniques can be successfully applied to improve the efficiency of the winemaking process.