There are a number of possible ways that the interaction between krill and its marine mammal predators can be viewed. A decision to manage is in part based on the belief that marine mammals can affect the yield of fisheries and the fisheries, as their food supply can affect the dynamics of marine mammals. The problems of interpreting the precise scientific meaning of the management criteria of Article II of the CCAMLR Convention are discussed. The likely behaviour of different components of the ecosystem is considered in the light of the following simple theoretical models: krill-whales, krill-whales-seals, krill-cephalopods­sperm whales. Practical difficulties of interpreting the current status of marine mammals from existing data are considered with special reference to possible competitive release resulting from the depletion of baleen whales. The different ways to acquire necessary information for management are mentioned including statistically designed perturbation experiments and monitoring of indicator species. Three classes of models can be used to assess mammal fisheries interactions. Theoretical models give guidance as to what sort of phenomena to look for, strategic simulation models indicate how to look for them and estimation models are the tools which summarise the observations for practical purposes. Such a process can be recursive. It is emphasised that modelling cannot substitute for experimentally rigorous observation and conversely, unguided observation provides only data, not insight.