Sea‐surface temperature (SST) is a key driver for various interactions and feedbacks between components of the Earth System and can control local weather and climate. The formation of marine fog, for example, can be sensitive to small changes in SST at a scale of a few kilometres. As a contribution to understanding processes at the interface between air and sea, this article discusses results from a state‐of‐the‐art fully coupled regional atmosphere–land–ocean–wave prediction system for the UK at km scale. This study focuses on the impact of the changes in surface forcing resulting from coupling SST in the marine boundary layer and formation of summertime coastal fog over the North Sea. A study from July 2013 provided a good case to evaluate the role of SST in fog evolution. The benefit of an evolving SST in the coupled simulation is shown in capturing a warming trend in observed SST over the five‐day case study period, with a root‐mean‐square error (RMSE) against in situ observations of 1.1 K. In contrast, in uncoupled atmosphere‐only simulations, the initial‐condition SST is persisted for the duration of the case, as is more typical in current operational numerical weather prediction (NWP). In the uncoupled simulations, a cold bias develops over the modelling period and the RMSE against observed SST is 2.4 K. The impact of coupling is shown to propagate into the overlying marine boundary layer and therefore affect the formation of coastal fog. Increased heat flux from a relatively warmer sea surface in the coupled simulations led to near‐surface atmospheric instability, hampering stratus lowering and destroying the fog‐promoting inversion layer. This significantly reduced fog fractions in selected regions. The value of model coupling was assessed by comparing coupled and uncoupled simulations initialized at different times ahead of fog development.