In the cascade, internal variability is represented by the range between different realizations from the same climate model (Fig. 4) and for the 2046–2065 period this range is 0.5 million km 2 on average across models and scenarios (0.6 million km 2 across the full CMIP5 ensemble; also see Supplementary Information). Internal variability on even shorter timescales, shown in the cascade by the range of pentads of a single realization, is 1.4 million km 2 on average, and reaches up to 4.6 million km 2 . Variability in 5-year means is largest when the sea-ice extent reaches near ice-free levels. This pattern is most clearly shown using the CESM1 LE, in which variability increases as the sea-ice retreats, before dropping to close to zero when ice-free conditions are reached (Supplementary Fig. 6). For the 20 year mean sea-ice extents over 2046–2065, model uncertainty is the dominant term (CMIP5 range of 9.4 million km 2), followed by scenario uncertainty (1.3 million km 2) and then internal variability (0.6 million km 2). It is worth noting that for the sea-ice extent trends considered in the previous sections, inter-realization spreads were not much smaller than the inter-model spread, even for multi-decadal trends (see Supplementary Information). For the multi-decadal means of sea-ice extent considered here, inter-realization spread is however much smaller than inter-model spread. Nonetheless, within any single The Nordic Seas are highly sensitive to environmental change and have been extensively monitored and studied across a broad range of marine disciplines. For these reasons, the Nordic seas may serve as a pilot area for integrated policy development in response to ongoing climate change. T he northern high-latitude seas and their coastal waters are among the most sensitive to climate change on Earth. Salinity, temperature and oxygen gradients will become steeper, wind patterns will shift, and the rapid increase in atmospheric CO 2 will continue to acidify the ocean. The critical question — not only for scientists across all disciplines, but also for policymakers and society in general — is how the combination of all these stressors will impact the interdependent ecosystems as well as the social systems within this region. These seas of Norden 1 are defined here as the Norwegian, Barents, Greenland and Iceland seas, as well as the Baltic and the North seas together with the ocean areas connecting them. Recognizing that they are interconnected, not only with each other, but also with human well-being and health, is a critical step in creating a chart to navigate science and policy towards a common goal of sustainability. Collaboration across scientific disciplines, between science, model, internal climate variability can play a significant role in determining sea-ice extent on decadal timescales, and it plays an even more important role on shorter timescales. Conclusions When accounting for internal climate variability, observed and simulated September Arctic sea-ice extent trends over 1979–2013 are not inconsistent. Internal variability can also either mask or enhance human-induced changes for decades at a time. Thus, pauses in sea-ice loss, such as seen over the past eight years, are not surprising and are fully expected to occur from time to time. Additional single model large ensembles that capture this variability would be valuable for advancing our understanding. Further evaluating the physical processes responsible for decadal variability in sea-ice extent in both observations and simulations will also improve our ability to understand how sea-ice is likely to evolve in the next few years and decades. ❐