Orbital parameters are assumed to drive glacial/interglacial cycles. However, pre-Quaternary (>2.7 myr ago) glacial cycles do not exhibit distinctive glacial amplitudes in the geological record. An atmosphere-ocean general circulation model (GCM) with dynamic vegetation is used for an extensive suite of studies, comprising changes in obliquity and precession, differential atmospheric CO2 values and climate background states representative for Pre-industrial, and Tortonian (11–7 myr), a relatively warm climate state prior Northern Hemisphere glaciation, is set up to test the sensitivity of climate. The model results show that the magnitude of impact of external insolation forcing is most dependent on atmospheric CO2 values and, to a lesser degree, on the background climate. Depending on the actual state of the climate system, sea-ice has been identified as a key regulator for mediating orbital forcing into a nonlinear climate response at high latitudes. This high latitude feedback has potential implications for Cenozoic hothouse climate as well as glacial inception. GCMs are tested and challenged by the ability to reproduce paleoclimate key intervals. In order to account for climate changes associated with soil dynamics, a soil scheme is developed, which is asynchronously coupled to a state-of-the-art atmosphere-ocean GCM with dynamic vegetation. The scheme is tested for conditions representative of a warmer (mid-Holocene, 6 kyr before present, BP) and colder (Last Glacial Maximum, 21 kyr BP) than pre-industrial climate. For these different climates the computed change in considered physical soil properties (i.e. albedo, total water holding field capacity, and texture) leads globally to an amplification of climate anomalies. Especially regions like the transition zones between desert/savannah and taiga/tundra, exhibit an increased response as a result of the modified soil treatment. In comparison to earlier studies, the inclusion of the soil feedback pushes the model simulations towards the warmer end in the range of mid-Holocene studies and beyond current model estimates of global cooling during the Last Glacial Maximum (LGM). The main impact of the interactive soil scheme on the climate response is governed via positive feedbacks, including vegetation dynamics, snow, sea-ice, local water recycling, which might amplify forcing factors ranging from orbital to tectonic timescales. Due to the lack of spatially and temporally sufficiently resolved reconstructions, the extent, thickness and drift patterns of sea-ice and icebergs in the glacial Arctic remains constrained. Earlier studies are contradictory and propose either a cessation of the marine cryosphere, or an ice drift system operating like present-day. Here, the marine Arctic cryosphere during the LGM is examined using a high-resolution, regional ocean-sea-ice model. Whereas in the modern western Arctic Basin sea-ice can circulate in the Beaufort Gyre for decades, glacial model studies present an extreme shortcut of sea-ice drift. The results show a clockwise sea-ice drift in the western Arctic Basin that merges into a direct trans-Arctic path towards Fram Strait. This is consistent with dated ice plough marks on the seafloor which show the orientation of iceberg drift in this direction. Also ice-transported iron-oxide grains deposited in Fram Strait, can be matched by their chemical composition to similar grains found in potential sources from the entire circum-Arctic. The model results indicate that the pattern of Arctic sea-ice drift during the LGM is established by wind fields and seems to be a general feature of the glacial ocean. Contradicting to former proxy reconstructions the model results do not indicate a cessation in ice drift during the LGM.