General principles of the USSR krill fishery such as the location of exploited fishing areas and the seasonal regime of their exploitation are considered. Using data obtained by the scouting vessel Globus engaged in regular krill fishery, it is shown that the catch-per-haul variables are associated with the fishing regime of the vessel rather than with krill abundance in a certain area. During preparations for regular fishing operations very short hauls (under 15 minutes) are practised. Such fishing practice, together with substantial fluctuations of catches during scouting operations often results in yields which do not correspond to the actual biomass of krill in the place in question. In both cases particular diurnal and long-term behaviour patterns have an impact. All these factors limit the extent to which CPUE can be used in simulation studies of krill distribution and stock assessment. A standard large-scale multi-disciplinary survey, followed by processing of the data obtained using diverse methods may be viewed as a better instrument for studies.