The winter of 1962/63 was the coldest in the UK in over a century while the mildest winter occurred in 1988/89. For countries to be resilient against the impacts of large weather variations in the future, it is important to understand the likelihood of seeing such extreme fluctuations in addition to future climate change. In Europe and North America, these fluctuations are related to a combination of year-to-year variability and low-frequency variability of the North Atlantic Oscillation (NAO). The NAO is now a significant source of predictability for seasonal forecasts in these regions, however the signal-to-noise ratio of the ensemble mean to total variability in these ensemble predictions has been shown to be anomalously small, which means the real world is more predictable than our climate models suggest. Here we provide a new evaluation of the ability of climate models to reproduce longer-term variability and extreme trends like those seen between the 1960s and 1990s, with a focus on the NAO. We also investigate relationships with other large scale changes such as the reduction in Arctic Sea Ice over recent decades. (PAMIP section: Research funded as part of APPLICATE project, Grant number 727862)