Bottle incubations, during which the activity and growth of prokaryotes is monitored during several days, are frequently carried out to study functional aspects of marine prokaryotic assemblages. These experiments will relate directly to in situ activities if all populations grow harmonically during the incubation. We tested whether this was the case by analyzing the composition of bacterial assemblages at the beginning and at the end of the incubation by denaturing gradient gel electrophoresis. Five experiments were done in different Antarctic regions. Bacterial assemblages north and south of the Polar Front were very different. In all cases, the final assemblages were very different from the initial ones, and these changes were often accompanied by a significant decrease of diversity indices. Our experiments included treatments with different temperature and organic matter amendments. Whereas the increase in temperature tested had a minor effect on prokaryotic growth rate and specific composition, the addition of organic matter strongly stimulated growth rate and selected a particular bacterial assemblage in some experiments but not in others. A significant component of bacterial assemblages from waters south of the Polar Front appeared to be Polaribacter franzmannii, a gas vacuolated bacterium of the CytophagaߚFlavobacteriumߚBacteroides group that was originally isolated from Antarctic sea ice. This phylotype was enriched and dominated in almost all final assemblages. Our results indicate that longߚterm bottle incubations mostly measure the activity of a few opportunistic bacteria and not that of the original assemblage. This should be taken into account if data obtained in these experiments are used for balancing whole ecosystem carbon budgets and to derive biogeochemical conclusions.