Mid-ocean ridges are divergent plate boundaries, where the seafloor spreads apart and new oceanic crust is formed. Yet, there are important differences between these individual ridges, which seem to be linked to the rate with which they spread apart. At full spreading rates of < 20 mm/yr, these ridges are called ’ultraslow’ spreading, and their appearance is drastically different from ridges which spread faster. Generally, ultraslow spreading ridges have a very rugged appearance, with steep rift flanks which contain numerous normal faults, and discontinuous volcanic activity in space and time at discrete volcanic centres. Magma supply is thought to be extremely limited, as mantle flow models infer a reduced melt production owing to the greater conductive cooling compared to faster spreading ridges. Crustal thickness as well is thought to be affected by the greater conductive heat loss of the ascending mantle, resulting in thick crust at volcanic centres, and only thin or absent crust in between. However, the processes which cause this are still only poorly understood. Ultraslow spreading ridges consist of alternating magmatic and amagmatic segments, each emphasizing a primary accretionary mode. Magmatic segments are robustly magmatic, with basalts exposed on the seafloor and axial volcanic ridges. In contrast, at amagmatic segments volcanic activity seems to be almost absent, and plate divergence is accommodated by the uplift of mantle horst blocks. Predominantly peridotites are dredged from the seafloor, interrupted by isolated centres of volcanic activity. Due to the remote locations of the two main representatives of this spreading class (Arctic Ridge System, Southwest Indian Ridge), only few surveys have so far been conducted. Gakkel Ridge, as part of the Arctic Ridge System, lies in the perennial ice covered Arctic Ocean far from any land, and the Southwest Indian Ridge is located between South Africa and Antarctica in latitudes with frequent storms. Both environmental conditions inhibit seismic surveys by making the loss of instruments likely, therefore so far seismicity studies of local earthquakes in particular are sorely missing. Yet, these kind of surveys are necessary to constrain information about the mechanical and thermal state of the lithosphere. In 1999, an exceptional earthquake sequence was teleseismically registered. This clustered seismicity was unusual in several aspects: The magnitudes involved (up to mb = 5.2), the number of events registered (252 earthquakes) and the duration of activity (9 months). It originated at 85�°E/85°�N at a volcanic centre in the eastern part of Gakkel Ridge where the spreading rate is about 10.2 mm/yr. The major part of this PhD-thesis deals with the investigation of this volcanic centre, first through the teleseismic sequence dataset and second through analysis of a local dataset gathered in 2007. This was published in two scientific publications. For any kind of analysis of the earthquake sequence of 1999 at 85°�E/8°5�N, reliable earthquake locations were needed. In the first part of my PhD-thesis, I analyzed this entire teleseismic earthquake sequence, first relocating the sequence, and second placing the obtained earthquake locations in a geological context. I relocated the entire teleseismic earthquake sequence with three different localization algorithms (NonLinLoc, Hyposat and Mlocate), While studying parameters which may influence the resulting locations systematically, I compiled a sub-dataset of reliably located events. Yet, locations from the different algorithms sometimes do not even match within their error ellipses. Thus, the choice of location algorithm has a critical influence on the use of a dataset which lacks nearby recording stations, and has to be made carefully. For absolute single-event localization, I preferred earthquake locations calculated by the algorithm NonLinLoc, which were placed closest to the epicentres calculated by the relative algorithm Mlocate. Both programs placed their epicentres farthest from locations calculated by the algorithm Hyposat. A temporal analysis of the earthquake sequence inferred three phases of activity, each with its own distinct character of seismicity. I visualized the most probable centre of earthquake activity in each phase and interpreted a complex interplay of tectonic and magmatic processes. In the first phase of seismicity the crust breaks either accompanying or enabling magmatic intrusion, followed by the second phase of seismicity during which the area producing seismicity got larger. The third phase of seismicity probably resulted from a post-intrusion adjustment of the stress field, or a transition to an effusive stage of volcanism. The modified Omori-Law describes the declining aftershock rate of tectonic mainshockaftershock sequences with succeeding time. It can thus be used to determine, if an observed seismicity sequence is tectonic in origin. I extensively researched this relationship for the second part of my PhD-thesis, and tried to match the modified Omori-Law to all suitable seismicity clusters at ultraslow spreading ridges. Yet, an exclusively tectonic origin could not be proved and therefore, the origin of these earthquake clusters most likely was influenced by magmatic interaction. Figuratively returning to the 85�°E/85°�N volcanic centre at Gakkel Ridge, I extensively analyzed a 16-day local dataset which had been collected in 2007, using seismometer arrays installed on ice floes which drifted over the survey area. This dataset shows the seismological aftermath of the spreading episode of 1999 at Gakkel Ridge, seven years after the major seismic activity. Due to the drift of the ice-floes, each earthquake was recorded at different station locations. This resulted in a sufficient ray coverage, suitable for the first local earthquake tomography ever done at an ultraslow spreading centre and this analysis formed a major third part of my PhD-thesis. I picked over 300 seismic events by hand, compiled a local velocity model from confidentlylocated hypocentres, located 248 hypocentres with this local velocity model, and generated a local earthquake tomographic model. The results infer a Moho at 7 km below seafloor, which implies an exceptionally thick crust. Deep hypocentres point to a cold lithosphere, in contrast to existing thermal models. An area of slower seismic velocities crosses the rift valley at the supposed site of volcanic activity in 1999. I interpreted the observed velocity anomalies to predominantly stem from the relaxation of thermal stresses following a recent intrusion. This PhD project gives several important insights into the seismological processes that accompany tectono-magmatic interactions at ultraslow spreading mid-ocean ridges. It culminated in the first local earthquake tomography ever done at a volcanic spreading centre of a magmatic segment at such a ridge. The unique results show that theoretical thermal models may have to be further refined, to accurately reflect spreading processes at ultraslow spreading ridges.