An operational management procedure for krill (Euphausia superba) in Subareas 48.1, 48.2 and 48.3 requires a basis for the assessment of resource status, and an algorithm for specifying the levels of regulatory mechanisms (e.g., a catch control law) that depends on the results of the assessment. The development and selection of a procedure requires a basis for the simulation testing of procedures, and an operational definition of CCAMLR Article II to provide criteria against which to assess procedure performance. Suggestions are made under each of these headings. Assessment of resource status is provided by the CPUE ‘Composite Index’ proposed by the Workshop on the Krill CPUE Simulation Study. Annual TACs are restricted to an initial ceiling (C_c) for a five year period, with a reference CPUE level (CPUE_ref) calculated as the average CPUE over that time. Thereafter TACs may increase by C_r% per annum. However, this increase may be suspended or reversed in any year, depending on how many of the previous three years' CPUE values fall below a target level of 0.75 CPUE_ref. An operating model of krill dynamics in the region is developed for simulation testing purposes. A provisional operational interpretation of Article II is proposed: the primary objective is to prevent the expected lowest biomass of krill over a 20-year harvesting period from falling below 60% of its average unexploited level; subject to this constraint, accumulated catches should be as large as possible without substantial associated probability that TAC reductions may prove necessary during the 20-year period considered. Simulation tests, including one particular test of robustness to the assumptions of the operating model, are carried out to illustrate the overall process proposed; for this illustrative exercise, the choice of catch control law parameters would probably lie between (C_c=1 million tonnes; c_r=15%) and (C_c=2 million tonnes; c_r=10%). Suggestions for proceeding with further investigations of possible operational management procedures are made. It is proposed that possible alternative suggestions for such procedures should be made in a similar fashion to that set out in the paper. Suggestions by others for alternative forms and parameter values (or their probable ranges) for the krill dynamics operating model used for testing procedures are encouraged.