Long runs (hundreds of years) of the global climate models and high-resolution climate proxy data sets provide an opportunity to study multi-decadal natural variability of the coupled atmospheric and oceanic circulation. The Nordic seas and the Arctic are regions were such a natural variability is expected to have large amplitude and regularity. Indeed, quasi-regular changes with the period of 60-90 years have been found in analysis of long control climate runs of several models. This model multi-decadal variability reminds the quasi-regular changes revealed in the ocean sediment and tree ring data in the area. The dynamical mechanism of the multi-decadal variability in the Atlantic sector of the Arctic was first proposed by J. Bjerknes (1964) who linked it to the compensation between the polarward ocean and atmospheric heat transport modulated by the extent of the sea ice cover. In the previous bilateral project (196178/S30: Detailed studies of the Bjerknes compensation mechanism), the researcher at NERS C and IARC have found strong variability caused by the Bjerknes compensation mechanism in the 600 yr simulations of the Bergen Climate Model. The geographical and temporal pattern of this variability have been studied and compared with the published analy sis of other models. This project continues the research in the context of climate historical data and future projections. Its goal is to establish correspondence between the multi-decadal variability found in the climate models and the variability derive d from Holocene paleo-proxy data sets. New and longer (up to 3000 years) runs of the climate models will be processed to improve the statistical significance of the variability patterns and dependences.