The ethanol production process of a Swedish alcohol production plant was dominated by Dekkera bruxellensis and Lactobacillus vini, with a high number of lactic acid bacteria. The product quality, process productivity, and stability were high; thus, D. bruxellensis and L. vini can be regarded as commercial ethanol production organisms.We analyzed the population dynamics of yeasts and lactic acid bacteria (LAB) in a Swedish ethanol production plant. The production process runs as a continuous fermentation with recirculation of the yeasts. The substrate for fermentation is formed from wheat starch. The material is first liquefied with ␣-glucoamylase at 90°C and then further degraded by ␣-glucosidase at 60°C to release fermentable sugars. By this procedure, about 96% of the starch is degraded to fermentable sugar. The glucose concentration in the fermentor is always below 0.1 g/liter because of the substrate-limited continuouscultivation method (B. Johansson, personal communication). The process is started by mixing 1 ton of baker's yeast (Jästbolaget, Sollentuna, Sweden; cell viability, Ͼ90%) with the substrate in the fermentor (fermentor size, about 100 m 3 ). According to observations of the staff, it usually takes up to 3 weeks until stable fermentation is obtained; thereafter, the process can run stably for 2 or even more years. During the first 3 weeks, the process is prone to infections and stuck fermentation. After stabilization, the staff noticed a change in the cell shape of the production yeast, which was regarded by them either as a physiological adaptation to the harsh conditions in the fermentor (high cell density, sugar and oxygen limitation, a pH of about 3.5, and a temperature of 35°C) or as a selected genetic variant of the inoculated baker's yeast. Our intention was to analyze this genetic variant but also to investigate the role of potentially contaminating LAB.Samples were taken in January, March, and July 2006 from the fermentor, which had been running stably since July 2005. Appropriate dilutions of the fermentation broth were spread onto plates selective for either yeasts (malt extract agar containing 20 g/liter malt extract and 0.1 g/liter chloramphenicol) or LAB (de Man-Rogosa-Sharp medium containing 0.1 g/liter Delvocid). Surprisingly, during the first two samplings the number of LAB was very high, constituting about 70% of the total cell number (yeast plus LAB). At the last sampling, this relation had changed and the LAB were less than 10% of the total cell number (Table 1). However, these changes did not influence the ethanol concentration in the fermentor (B. Johansson, personal communication). The quantifications are based on CFU determinations on the selective media described above. It was not possible to perform microscopic counts because of the high number of particles in the industrial medium that were difficult to distinguish from microbial cells. From every sample, 20 yeast and 20 bacterial colonies were randomly chosen for PCR fingerprint analysis. For a comparison, we also anal...