Stratified random acoustic surveys, each comprising 10 parallel transects, were conducted within two 80 x 100 km areas over the shelf-break to the north-east and north-west of South Georgia during December/January 1996/97. Netting suggested that the mean lengths of krill within the eastern and western survey boxes were 36.1 mm and 45.4 mm respectively. These lengths were used to derive box-specific krill target strength (TS) values of -38.89 and -38.59 dB kg-1. Acoustic signals believed to be attributable to krill were identified on the basis of the difference between mean volume backscattering strength at 120 and 38kHz (mvbsl20 kHz - mvbs38kHz between 2 and 12 dB indicative of krill). Mean volume backscattering strengths at 120 kHz were scaled by the appropriate TS value to determine mean krill biomass for each transect. Transect values were used to derive weighted mean (and variance) estimates of krill biomass within each survey area. The estimates were 58.28 g m-2 (56.31) and 25.17 g m-2 (18.44) for the eastern and western boxes respectively. These mean density values were similar to those obtained during the n10st recent previous surveys of krill biomass (January 1996) in these areas, but substantially higher than estimates derived from surveys there in January 1994.