Understanding the flow of ice is essential to predict the contribution of the polar ice sheets to global mean sea level rise in the next decades and centuries. During this PhD project, the recrystallization and deformation mechanisms that govern the flow of ice were studied along the length of the North Greenland Eemian Ice Drilling (NEEM) ice core in northwest Greenland. Two methods were used during this study: (i) cryogenic electron backscatter diffraction (cryo-EBSD) in combination with (polarized) light microscopy and (ii) flow law modelling using two different flow laws for ice constrained by the actual temperature and grain size data from the NEEM ice core. The NEEM ice core was divided up into three depth intervals originating from different climatic stages that differ strongly in terms of impurity content, microstructure, deformation mode and temperature: the Holocene ice (01419 m of depth), the glacial ice (1419-2207 m of depth) and the Eemian-glacial facies (2207-2540 m of depth). Microstructures indicate that the Holocene ice deforms by the easy slip system (crystallographic basal slip) accommodated by the harder slip systems (non-basal slip), also known as dislocation creep, and by recovery via strain induced boundary migration (SIBM), which removes dislocations and stress concentrations and allows further deformation to occur. The amount of non-basal slip that is activated is controlled by the extent of SIBM. The dominant recrystallization mechanisms in the Holocene ice are SIBM, bulging recrystallization and grain dissection in total leading to dynamic grain growth, with a contribution from normal grain growth in the upper 250 m. The strain rate variability with depth in the Holocene ice, estimated by flow law modelling, is low. In contrast, the strain rate variability is relatively high in the glacial ice as a result of variability in grain boundary sliding (GBS) with depth that accommodates basal slip (GBS-limited creep). Grain boundary sliding in the glacial ice is particularly strong in fine grained sub-horizontal bands which contain many aligned grain boundaries. Subgrain boundaries (SGBs) form ahead of the aligned grain boundaries in these fine grained subhorizontal bands and when a misorientation angle of 5.0°-6.0° is reached, the SGB has rotated into a sliding boundary. Rotation recrystallization is more prominent, while SIBM is less important in the glacial ice compared to the Holocene ice. The ice in the Eemian-glacial facies, which is affected by premelting along the grain boundaries, alternates between relatively fine grained glacial ice with a single maximum crystallographic preferred orientation (CPO) and very coarse grained Eemian ice with a partial girdle CPO. Due to the difference in grain size and CPO, it is argued that the glacial ice in the Eemian-glacial facies deforms almost entirely by GBS-limited creep in simple shear and at high strain rates, while the Eemian ice in the Eemian-glacial facies deforms at much lower strain rates in coaxial deformation with a roughly equal contribution of GBS-limited creep and dislocation creep to bulk strain rate. The large difference in microstructure, and consequently viscosity, between impurity-rich glacial ice and impurity-depleted interglacial ice in the premelting layer (262K<T<273K) of polar ice sheets can have important consequences for ice dynamics close to the bedrock, which provides the major contribution of the horizontal movement of ice towards the oceans. Glen’s flow law (Glen, 1955; Paterson, 1994), which is independent of grain size, predicts a higher strain rate along the NEEM ice core than the grain size sensitive composite flow law of Goldsby and Kohlstedt (1997, 2001) for which the flow law parameters of the dislocation creep mechanism were modified during this PhD project. Only a grain size sensitive flow law can predict the variability in strain rate resulting from a variability in grain size with depth, which is expected to be large in the NEEM ice core.