Concerns regarding the acoustic quality of habitats have increased over the past years. Within the EU Marine Strategy Framework Directive, this has led to the proposal of environmental status indicators for underwater noise levels. One indicator, based on 1/3- octave band levels at 63- and 125-Hz, intends to monitor anthropogenic noise, particularly from shipping. In this context, ambient noise is defined as the acoustic energy not assignable to a specific source. We applied this indicator to passive acoustic underwater recordings from the Southern Ocean, which represents an environment relatively void of anthropogenic noise. To collect multi-year passive acoustic data, two autonomous underwater recorders were deployed in the Weddell Sea at 0°W/E 66°S and 0°W/E 69°S between March 2008 and December 2010. The analysis of the data revealed three caveats in the current implementation of this indicator: 1.) Usually, band noise calculations are based on temporal averages over all available data at the temporal resolution desired. However, this approach will bias towards transient, single loud and nearby acoustic events rather than representing the typical ambient noise. We propose to calculate ambient noise levels on the basis of the quietest 10 s each e.g. 5-min recording. 2.) Ambient noise levels in the 125-Hz band were strongly influenced by biotic sources. We propose that definitions of noise bands suitable for monitoring of anthropogenic noise need to be tuned as to avoid region-specific biotic interference. 3.) Our recordings reveal a substantial seasonal variability of ambient noise levels (102- 115 dBrms re 1 μPa, broadband sound pressure level 5th and 95th percentile), even if selected outside biotic bands, primarily due to sea state and ice cover. This implies that detecting gradual changes in ambient noise (e.g. the frequently reported 0.3dB/a) on the basis of annual averages is rather challenging if statistical significance is required.