A 3-d baroclinic coupled ice–ocean model, applied to the connected marginal seas, North Sea and Baltic Sea, was used to investigate the seasonal cycle of both heat content of the water column and atmosphere–ocean heat exchange throughout the seasonal cycle. Case studies were carried out to investigate, quantify and inter-compare the intra-annual sensitivity of the thermal state of both marginal seas in response to changes in wind forcing, air temperature and fresh water runoff. The prescribed changes in model forcing were well within the range of the observed variability. A simulation for a representative reference case (1984–84) served to quantify predicted anomalies. Reducing the fresh water runoV for both seas by 30% resulted in a surprisingly small response in the heat content which was one order of magnitude smaller as compared to the applied change in wind forcing. A reduction of the air temperature by 2 °C caused a decrease of the heat content throughout the seasonal cycle in the order of 30%. In contrast to a change in air temperature a reduction of 30% in wind stress yielded distinct seasonal differences in the oceanic response. The most significant wind induced changes occurred during autumn and winter in the Baltic Sea and in the North Sea. A reduced wind forcing led to a larger oceanic heat content in winter as a consequence of a reduced winter convection and an intensification of the winter thermocline in the freshwater dominated Baltic. In the Baltic Proper, with its perennial thermo-haline stratification, predicted temperature changes of intermediate waters were several times higher than sea surface temperature changes. Compared with the North Sea, the Baltic showed a much higher sensitivity in response of the heat content to changes in the wind forcing. However, the opposite is true for the heat flux from the water to the atmosphere during the cooling period. Here the sensitivity of the North Sea is much higher than that of the Baltic Sea. This is caused by the fact that advective heat flux changes and atmospheric heat flux changes are working in the same direction in the Baltic Sea and acting in opposite direction in the North Sea. Results of this sensitivity study suggest that future studies on an inter-annual sensitivity should be conducted with a coupled atmosphere–ocean model.