Abstract. A 1-D biogeochemical/physical model of marine systems has been applied to study the oxygen cycle in four stations of the different sub-basins of the Baltic Sea, namely, in Gotland Deep, Bornholm, Arkona and Fladen. The model consists of biogeochemical model of Neumann et al. (2002) coupled with the 1-D General Ocean Turbulence Model (GOTM). The model has been forced with meteorological data from the ECMWF reanalysis project for the period 1998–2003, producing a 6-year hindcast validated with datasets from the Baltic Environmental Database (BED) for the same period. The vertical profiles of temperature and salinity are relaxed towards both profiles provided by 3-D simulations of General Estuarine Turbulent Model (GETM) and observed profiles from BED. Modifications in the parameterisation of the air/sea oxygen fluxes have led to significant improvement of the model results in the surface and intermediate water levels. The largest mismatch with observation is found in simulating the oxygen dynamics in the Baltic Sea bottom waters. The model results demonstrate the good capability of the model to predict the time-evolution of the physical and biogeochemical variables at all different stations. Comparative analysis of the modelled oxygen concentrations with respect to the observation data is performed to distinguish the relative importance of several factors on the seasonal, interannual and long-term variations of oxygen. It is found that the natural physical factors, like the magnitude of the vertical turbulent mixing, wind speed, the variation in temperature and salinity field are the major factors controlling the oxygen dynamics in the Baltic Sea. The influence of limiting nutrients is less pronounced, at least under the nutrient flux parameterisation assumed in the model.