OBS are meanwhile routinely used in the world’s oceans to record local and teleseismic earthquakes for periods of up to 2 years. They have been deployed on mid-ocean ridges, in subduction trenches and on abyssal plains. However, OBS deployments for passive seismic studies in the stormy Southern Ocean and the ice-covered Arctic Ocean have so far not been attempted due to logistic challenges arising from rough sea and perennial ice cover. We have deployed and recovered for the first time an OBS network on the Southwest Indian Ridge in the stormy “Furious Fifties” during Polarstern cruises ANT-XXIX/2 and ANT-XXIX/8. In addition, during Polarstern cruise PS86 to western Gakkel Ridge in July 2014, we performed a test deployment of a single OBS at 80%-90% sea ice cover. We report here on our experiences and special approaches used to deal with wind, ice and waves. For local networks, the OBS position at the seafloor is critical to know yet hard to determine to within 200 m or less. Acoustic ranging of the OBS proved to be inconclusive or even impossible in case of Polarstern whose noise prevents receiving the acoustic signals. Active seismic profiling allowed us to determine the real OBS position which turned out to differ significantly from deployment and recovery position. During the interdisciplinary cruise ANT-XXIX/8, we stayed in the survey area for an extended time period and could wait for suitable conditions for safe OBS recovery. For future surveys in stormy areas and for routine use in sea ice, however, tracking of the OBS during ascent is necessary to exactly know the surfacing position warranting safe recovery. For our test deployment in sea ice, we conceived a head buoy that is fixed to the OBS and only released upon recovery. This avoids entangling of the buoy and the winch cable. We added additional weight to the OBS and winched it to the seafloor. A Posidonia buoy on a 100 m line was attached to the OBS. It hangs down from the OBS once it ascends and its absolute position can readily be tracked during ascent until the OBS surfaces even if it gets stuck under an ice floe. In our test, the Posidonia buoy imploded after reaching the seafloor, but the OBS could be ranged acoustically during ascent and was safely recovered in a pond of open water. For future deployments in stormy waters and sea ice, a light-weight Posidonia transponder that can be set to sleep mode would be a desirable addition to the DEPAS instruments. A first inspection of the data showed that the ice-covered Arctic Ocean has very low noise levels due to the lack of ocean swell, while especially the Southern Ocean displays considerable noise at periods of 0.5s-8s.