An initial attempt is made to develop the model1ing framework suggested by the Joint Meeting of the CCAMLR's WG-Kril1 and WG-CEMP in 1992 to address this issue. First. estimates are made of the parameters of predator survival rates as functions of krill abundance, by considering a krill dynamics model incorporating recruitment fluctuations together with information on adult survival and breeding success patterns for certain krill predator species. A "one-way" interaction model is developed, in which krill abundance fluctuations impact the predator population, but not vice versa. Computations based on this model indicate that variability in the annual recruitment of krill results in predator populations being less resilient to krill harvesting than deterministic evaluations would suggest. However, the analyses also raise a number of questions about the proper interpretation of the available predator population dynamics information in the context of the models developed, and about the model1ing of the predator survival rates as functions of krill abundance. It is suggested that these questions merit discussion at the forthcoming WG-Kril1 and WG-CEMP meetings. A formalism for a "two-way" interaction model (including also the effect of differing predator consumption levels on krill) is developed, but computations based on this approach are deferred pending clarification of the Questions raised above at the WG-Krill and WG-CEMP meetings.