Acoustic surveys designed to assess krill abundance are costly in time and money, so the opportunity was taken in 1992 to produce a krill biomass estimate as a by-product of a fish stock-assessment surveyin CCAMLR Subarea 48.3. Acoustic transects were run between trawl stations using a sounder operating at 38 kHz and 120 kHz. The results have been analysed for all straight sections of track when the ship's speed exceeded 7 knots and when the ship was over the shelf around the main island of South Georgia. The appearances of echoes on echo charts were used to decide which echoes to include in krill estimates. The results were partitioned by depth to remove deep echoes that were thought to be mostly due to other scatterers. A threshold at one frequency was used to remove noise and any echoes too weak to be separated from the background. The remaining fraction was further subdivided using the ratio of backscattering strength at the two frequencies into a) echoes from krill-sized scatterers and smaller and b) echoes from larger scatterers. The data in the krill-size subset has been converted to density estimates, these are presented for day and for night sections of survey and at locations mapped around the island. An overall mean for daytime transects was 95 g m-2.