Genetic improvement of industrial yeast strains is restricted by the availability of selectable transformation markers. Antibiotic resistance markers have to be avoided for public health reasons, while auxotrophy markers are generally not useful for wine yeast strain transformation because most industrial Saccharomyces cerevisiae strains are prototrophic. For this work, we performed a comparative study of the usefulness of two alternative dominant selectable markers in both episomic and centromeric plasmids. Even though the selection for sulfite resistance conferred by FZF1-4 resulted in a larger number of transformants for a laboratory strain, the p-fluoro-DL-phenylalanine resistance conferred by ARO4-OFP resulted in a more suitable selection marker for all industrial strains tested. Both episomic and centromeric constructions carrying this marker resulted in transformation frequencies close to or above 10 3 transformants per g of DNA for the three wine yeast strains tested.During the past century, the selection of microbial strains has been one of the main sources of improvement for traditional fermentation processes, including the production of fermented foods such as wine or dairy products. For enology, the natural variability of Saccharomyces cerevisiae strains has been exploited by means of the isolation of wild yeast strains from grapes, musts, or successful fermentations; characterization of the isolates; and selection of the most suitable strains for each particular wine-making process or production area. Classical genetic methods such as the isolation of spontaneous or induced mutants or sexual crosses have been seldom used for the improvement of S. cerevisiae strains used for wine making (2,19,23,26,37), probably because of the difficulties in the design of selection procedures and because of the fact that, very often, the progeny of an excellent wine-making strain performs poorer than the parent strain (18). During the last 15 years, several attempts to construct genetically engineered yeast strains have been published, and very interesting improvements in the wine-making process or the quality of the wine obtained have been reported, including improved primary and secondary flavors, malic acid decarboxylation by yeast, increased resveratrol, lactic acid, or glycerol contents, and improved survival properties under technological conditions (6,11,16,17,20,27,32,34,38,43,46,47). However, because wild yeast strains are usually prototrophic, most of the strains obtained for these works contained antibiotic resistance genes which were used for selection of the recombinant strains and would be unacceptable for marketing. Puig et al. (35) published an interesting method for obtaining recombinant wine yeast strains that are free of antibiotic selection markers. In this method, a ura3⌬ strain is first constructed by sequential gene replacement of the wild-type URA3 gene, and the kanamycin resistance cassette used to this end is removed by homologous recombination. The resulting auxotrophic strain can then be ...