Despite the considerable attention that has been paid to bacterioplankton over recent decades, the dynamic of aquatic bacterial community structure is still poorly understood, and long-term studies are particularly lacking. Moreover, how the environment governs diversity patterns remains a key issue in aquatic microbial ecology. In this study, we used denaturing gradient gel electrophoresis of PCR-amplified partial 16S rRNA gene fragments and multivariable statistical approaches to explore the patterns of change in the free-living bacterial community in the mesotrophic and mono-meromictic Lake Bourget (France). A monthly sampling was conducted over two consecutive years (2007 and 2008) and at two different depths characterizing the epi- and hypolimnion of the lake (2 and 50 m, respectively). Temporal shifts in the bacterial community structure followed different patterns according to depth, and no seasonal reproducibility was recorded from 1 year to the next. Our results showed that the bacterial community structure displayed lower diversity at 2 m (22 bands) compared to 50 m (32 bands) and that bacterial community structure dynamics followed dissimilar trends between the two depths. At 2 m, five shifts in the bacterial community structure occurred, with the temporal scale varying between 2 and 8 months whereas, at 50 m, four shifts in the bacterial community structure took place at 50 m, with the temporal scale fluctuating between 3 and 13 months. More than 60% of the bacterial community structure variance was explained by seven variables at 2 m against eight at 50 m. Nutrients (PO(4)-P, NH(4)-N and NO(3)-N) and temperature were responsible for 49.6% of the variance at 2 m whereas these nutrients, with dissolved oxygen and chlorophyll a accounting for 59.6% of the variance at 50 m. Grazing by ciliates played also a critical role on the bacterial community structure at both depths. Our results suggest that the free-living bacterial community structure in the epi- and hypolimnion of Lake Bourget is mainly driven by combined, but differently weighted, top-down and bottom-up factors at 2 and 50 m.