Accurate estimation of krill biomass by acoustic techniques is dependent on a number of factors of which one of the most fundamental is accurate echosounder calibration. The Simrad EK500 scientific echosounder used aboard RRS James Clark Ross is calibrated regularly at South Georgia before and after krill surveys there, and exhibits temporal stability in system gain settings. Between Antarctic seasons this echosounder has also been calibrated in temperate European waters. Under these warmer conditions, calibrated gain settings differ markedly from those applied in the Antarctic, even after adjustments have been made to account for differences in sound speed between locations. Here we present results from multiple Antarctic and European calibrations which suggest that echosounder transducer performance is dependent on ambient water temperature. Highly significant differences in volume backscattering (S_v) and target strength (TS) transducer gains were detected at both 38 and 120 kHz between calibrations conducted at the two locations. At 120 kHz, the required S_v transducer gains at South Georgia (sea temperature at depth of transducer = 2.3^oC) were on average 1.4 dB less than in European waters (16.6^oC), and a similar trend was detected at 38 kHz. If European calibration parameters were to be employed during surveys around South Georgia, and no account were taken of the differences in gain settings, then integrated 120 kHz echo signals would be under-reported by 2.8 dB, leading in turn to an under-estimation of krill biomass by 52.5%. Every effort should therefore be made to ensure that echosounders are calibrated at temperatures as close as possible to those prevailing within the area in which surveys are conducted. In addition, the implications for biomass estimation of temperature variation across a survey area should be considered carefully