There was recognition that including coupling between atmosphere, sea ice, and ocean can improve forecast skill across a range of time scales. Coupling must therefore be considered across a range of scales appropriate to the time scale of the forecast, from planetary-scale dynamics to subgrid-scale processes in models. It was pointed out that better knowledge is needed as to the dominant causal direction of atmosphere-ocean-ice coupling and polar-midlatitude-tropical interactions and of their physical mechanisms, including the role of the stratosphere. In addition, the effects of model biases on projections for many of the phenomena discussed are still poorly understood. Participants pointed out that further work is required to elucidate the anthropogenic contribution to sea ice trends and to better understand the character of anthropogenic changes in the large-scale atmospheric circulation. Another key message emanating from the discussion was that Arctic-midlatitude interactions are clearly two way, which complicates attribution of cause and effect. For example, Arctic amplification is itself partially a response to lower-latitude forcing. Participants suggested that future research with respect to extreme events at high latitudes should be extended to oceanography, biology, avalanches, and visibility. Furthermore, there was a general agreement to enhance stakeholder involvement with respect to polar predictability. Last, workshop participants pinpointed the importance of process-based studies to make progress on research questions where insufficient observations and/or inadequate models hamper assessment of larger temporal- and spatial-scale processes.