Only with an improved understanding of natural climate variability confident estimations ofpossible climatic changes due to anthropogenic influences can be given. In view of the observedincrease of global mean surface air temperature since the beginning of the twentieth centuryinvestigations of natural climate variability have to concentrate on time scales of decades tocenturies. While the role of external factors including anthropogenic effects in causing climatechanges on these time scales have been investigated widely, the knowledge of the contributionof nonlinear atmospheric processes to climate variability is rather poor. This shortcoming existsas a result of the absence of statistically sufficient long atmospheric observational data seriesand as a result of the limited temporal and horizontal resolution of paleoclimate proxy data. Toaddress the problem of internal atmospheric low-frequency variability anyway long-termintegrations of appropriated models are one possible way to get hints on the real atmosphericbehaviour on decadal and interdecadal time scales.