Arctic sea ice has major impact on the global climate as it affects the energy budget of the Earth and its export from the Arctic towards the North Atlantic has direct influence on the global ocean circulation. The observed modern sea ice loss, mainly associated with anthropogenic greenhouse gas emissions, has raised the concern about the role and natural variability of sea ice and its relation to short term changes in the climate system. The reconstruction of past sea ice changes based on sediment cores and analyses of specific proxies provides vital knowledge about the natural variability of sea ice in pre-industrial times. Moreover it helps to understand the present changes and may improve estimates for future changes and consequences. Particular important areas for sea ice reconstructions are the shelves around Greenland. They connect the Arctic Ocean to the North Atlantic and underlie the major outflow of freshwater and sea ice from the Central Arctic. In this context, a specific interval of interest for climate reconstructions is the late Holocene, as it exhibits modern boundary conditions, i.e., oceanography and geography, and encompasses several well-known short-term climate events, e.g., the Medieval Warm Period and Little Ice Age. For past sea ice conditions and open water phytoplankton production, the application of specific biomarkers, i.e., highly branched isopenoids (IP25, HBI III) and sterols (brassicasterol and dinosterol) have proven as useful and reliable proxies. However, their applicability, especially in regard to HBI III, is mostly confirmed on regional scales so far. The PIP25 index, the ratio of IP25 to an open water phytoplankton biomarker, is used for a more quantitative sea ice reconstructions. Aim of this study is to confirm the IP25/PIP25 approach in surface sediments on an over-regional scale and for specific regions, i.e., the Baffin Bay, by comparing them to satellite-derived modern sea ice concentrations. Further the PIP25 approach is compared to other common microfossil methods, i.e., dinocysts and diatoms, for sea ice reconstructions. Following this, two well-dated cores from the East Greenland Shelf (PS2641-4/PS2641-5) and the West Greenland Shelf (MSM05/3-343310) covering the last 5.2 and 2.2. kyr BP, respectively, were analysed for their biomarker content. By analysing a nearly circum-Arctic surface sediment database that combines new and published data, the distribution of IP25 and specific open water phytoplankton sterols (brassicasterol, dinosterol) could be related to modern, satellite derived sea ice conditions in (sub-)Arctic regions. In regard to the phytoplankton marker HBI III, we find promising results in its distribution in relation to sea ice, however some regions remain unclear. The circum-Arctic distribution of the sea ice index PIP25, (i.e., PBIP25, PDIP25 and PIIIIP25) can be related to the modern spring sea ice distribution. However, the direct correlation to modern sea ice concentrations remains difficult for specific areas, i.e., the Central Arctic Ocean and the Russian shelves, which may be related to complex environmental conditions. A regional comparison of the PIP25 and MAT dinocyst reconstructions in the Baffin Bay revealed that both methods show a positive correlation with modern sea ice concentrations, with slightly higher correlations of the PIIIIP25 index. Further, diatom assemblages from the Baffin Bay, show highest correlations to modern spring sea ice, however the over regional application remains questionable. In regard to the paleo sea ice reconstructions we find specific variabilities on the East and West Greenland shelves. A mid- to late Holocene (last 5.2 kyr BP) sea ice reconstruction from the East Greenland Shelf revealed that IP25/PIP25 index based sea ice reconstructions do not reflect the wide-spread late Holocene Neoglacial cooling trend that follows the decreasing solar insolation pattern. This may be related to the strong influence of the polar East Greenland Current on the East Greenland Shelf and interactions with the adjacent fjord system throughout the studied time interval. However, several oscillations with increasing/decreasing sea ice concentrations that are linked to the known late Holocene climate cold/warm phases, i.e., the Roman Warm Period, Dark Ages Cold Period, Medieval Climate Anomaly and Little Ice Age, were revealed. The observed changes seem to be related to general ocean/atmosphere circulation changes, possibly related to North Atlantic Oscillation and Atlantic Multidecadal Oscillation regimes. On the West Greenland Shelf, at the mouth of Disko Bugt, however, biomarker sea ice reconstructions from the late Holocene indicate a gradual expansion of sea ice over the past 2.2 kyr following the Neoglacial cooling. Maximum sea ice conditions were reached during the Little Ice Age and we find evidence for the presence of a stable spring ice edge around 0.2 kyr BP. The Disko Bugt record revealed no other clear evidence for sea ice changes related to late Holocene climate events. However, superimposed on the general trend, a short-term oscillation in open water primary production and terrigenous input may be related to changes in the Atlantic Multidecadal Oscillation and solar activity as trigger mechanism. Within this thesis the nearly Arctic wide application of IP25 as sea ice proxy and the regional correlation of the PIP25 index with modern sea ice concentrations was further approved. The studies revealed substantial differences between East and West Greenland Shelf sea ice conditions during the Holocene. The studied location on the East Greenland Shelf was highly sensitive for small-scale Holocene climate events whereas the West Greenland Shelf reflected the general Neoglacial cooling trend. Further studies will improve the knowledge on the application of the biomarker sea ice reconstruction approach and the relationship of East and West Greenland sea ice development in late Holocene times.