Selecting optimal locations for mussel farming is vital for the optimization of production yield and for the minimization of environmental impact. Although predation by sea ducks may induce large stock losses and hence severe economic loss for mussel farmers, predation potential is rarely included in site-selection tools. In this paper we present a GIS-based spatial model predicting the potential of eider predation on blue mussel farms in Danish coastal waters. The model incorporates national survey data on eiders, as well as knowledge of eider behavior and habitat preferences, and was calibrated with predated/non-predated observations of eiders from 9 experimental mussel farms or test lines in Danish coastal waters. Except for 1 case study area, our model successfully confirmed a higher predation potential at test sites where predation had been observed. Our resulting predation potential map revealed potentials ranging from very low in inner parts of narrow estuaries to very high in more open coastal areas. Integration of the predation map into an existing site-selection tool showed that areas optimal for mussel growth were also associated with the highest modelled predation potential. Nonetheless, it was possible to identify areas having a very low potential of predation and only a 10% lower mussel production potential. These results underpin the potential for reducing production loss and increasing income by including predation potential in site-selection tools. In addition, the eider predation model can be used to identify and subsequently protect key foraging areas to support eider conservation.