The Finite Element Sea-ice Ocean Model (FESOM) is formulated on unstructured meshes and offers geometrical flexibility which is difficult to achieve on traditional structured grids. In this work, the performance of FESOM in the North Atlantic and Arctic Ocean on large time scales is evaluated in a hindcast experiment. A water-hosing experiment is also conducted to study the model sensitivity to increased freshwater input from Greenland Ice Sheet (GrIS) melting in a 0.1-Sv discharge rate scenario. The variability of the Atlantic Meridional Overturning Circulation (AMOC) in the hindcast experiment can be explained by the variability of the thermohaline forcing over deep convection sites. The model also reproduces realistic freshwater content variability and sea ice extent in the Arctic Ocean. The anomalous freshwater in the water-hosing experiment leads to significant changes in the ocean circulation and local dynamical sea level (DSL). The most pronounced DSL rise is in the northwest North Atlantic as shown in previous studies, and also in the Arctic Ocean. The released GrIS freshwater mainly remains in the North Atlantic, Arctic Ocean and the west South Atlantic after 120 model years. The pattern of ocean freshening is similar to that of the GrIS water distribution, but changes in ocean circulation also contribute to the ocean salinity change. The changes in Arctic and sub-Arctic sea level modify exchanges between the Arctic Ocean and subpolar seas, and hence the role of the Arctic Ocean in the global climate. Not only the strength of the AMOC, but also the strength of its decadal variability is notably reduced by the anomalous freshwater input. A comparison of FESOM with results from previous studies shows that FESOM can simulate past ocean state and the impact of increased GrIS melting well.