During the past two years, the AWI prepared a comprehensive, strategic assessment of the risk posed to marine mammals by the use of airguns for scientific, geophysical research in the Southern Ocean around Antarctica. The study attempts distinguishing between aspects of analysis (based on scientific knowledge and numerical calculations) and evaluation (based on a set of risk criteria and associated thresholds). The term assessment is used to describe the overall process, involving both analysis and evaluation. The analytical part commences with a synopsis of environmental (oceanographic and bathymetric) and operational characteristics from all seismic expeditions conducted in the Antarctic by the AWI. This resulted in a set of 4 basic environmental scenarios, which, combined with a set of 6 airgun configurations, were used to calculate single shot acoustic fields (SPL and SEL) by numerical (finite difference 2.5 D full waveform) modelling for a realistic ocean of 10x10 km dimension, resulting in a total of 24 different acoustic scenarios [Breitzke and Bohlen, 2010]. The current state of ecological knowledge was compiled for the 14 cetacean and 6 pinniped species to which the Antarctic represents an important habitat. Ecological and physiological information such as dive cycles and hearing curves are subsequently used to guide estimations of cumulative exposure levels and to develop mitigation measures. The identification of hazards including associated evaluation criteria and threshold levels provides the critical link between the sound propagation analysis and the risk evaluation. As currently no legally binding set of numerical threshold levels exists for this ocean region, we conducted a literature review of the pertinent literature. Three different risk categories were identified, for which a set of evaluation criteria was extracted from primarily three recent overview articles: (1) Southall et al., [2007], which provides numerical thresholds for the risk of direct, immediate injury; (2) Cox et al., [2006], from which a list of abetting factors was extracted for the risk of indirect, immediate damage, i.e. the so-called beaked whale scenario; and (3) The National Research Council [2005], which suggests a set of mostly qualitative evaluation criteria for the risk of biologically significant acoustic disturbance. By applying these criteria to the modelled acoustic fields (under the assumption of the ship following a straight course), critical exposure radii were calculated for single and multiple exposures. Finally, the resulting risk for individual animals and ensuing risks at the population level were evaluated by including information on species status and migratory behaviour, considering operational scenarios both with and without proposed mitigation measures in place. The analysis reveals that the risk for a marine mammal to incur direct, immediate injury from multiple exposures cannot be excluded in the immediate vicinity of the airgun clusters while indirect, immediate damage of an individual however appears rather unlikely. A risk of biologically significant acoustic disturbance, while negligible for juveniles and adults, cannot be excluded for the (merely hypothetical, hitherto unobserved) possibility of individual mother-calf pair separations. A possible manifestation of any of these risks however depends on a whale actually being within the respective acoustic range of the ship. This results in a negligible probability for risks at a population level, with the exception of a not to be excluded possibility of population level consequences for the Antarctic blue whale due to the abovementioned (hypothetical) possibility of mother/calf separations. The probability of this later impact is however estimated to be smaller than estimates of the natural mortality rate or of the PBR (possible biological removal) as used in other contexts. Noting the a) large existing gaps in the knowledge pertinent to this issue, and b) the fact that marine mammal behaviour is not fully predictable for an individual animal, it is unavoidable to base parts of this and other risk assessments on extrapolations of the current best knowledge, statistical descriptions of typical behaviour and even on educated guessing. When such steps had to be taken here, we here attempted to adhere to a conservative approach in our calculation and evaluation of contingent risks. The term conservative thereby stands for a selection of parameters or proxies, which chooses those that overestimate the risk while providing increased protection for the marine mammal. Nevertheless, with the risk evaluation being critically dependent on the thresholds used, it was no surprise that some selections (to be presented in detail at the meeting) made in this study were met with disagreement when discussed with various stakeholders. It is with these controversial issues, that the scientific communitys expertise and guidance preferably in form of peer reviewed publications - would be most helpful to further develop balanced and objective risk assessments acceptable to the majority of stakeholders.