The generalised yield model (GYM) was used by CCAMLR to establish the precautionary catch limit for the Antarctic krill (Euphausia superba) fishery. The current precautionary catch limit was based on supplying the GYM with a natural mortality rate of 0.8 and recruitment variability generated using a Beta distribution for proportional recruitment of krill. In this study, krill sampling data for empirical size frequencies were supplied to the GYM as the ‘vector of recruitments’ input option to simulate the population dynamics of krill around the Antarctic Peninsula (Subarea 48.1) along with increasing rates of natural mortality. The annual proportions of krill less than 36 mm in length to the total captured in net samples in four sampling areas of the Peninsula were used as proxies for recruitment variability. The variability of proportional recruitment in the CCAMLR study areas was similar to the variability in other krill studies and in the annual size distributions of krill in penguin diets. Simulations with either no fishing, or with fishing at the trigger level (lowest catches), at approximately half the precautionary catch limit (intermediate), or at the precautionary catch limit (highest) were conducted. As the values for natural mortality, recruitment variability and catch were increased, fewer of the scenarios were able to meet the CCAMLR decision rules. The higher precautionary level of catch was not obtainable while meeting CCAMLR decision criteria for at least two of the four recruitment vectors based on net sampling, regardless of how the specified parameters for recruitment and mortality were combined. Any substantial future increases in krill harvests in Area 48 beyond the trigger level require verification that the krill recruitment variability, natural mortality, and other parameters specified in the scenarios used to test management criteria, adequately represent the range of plausible values encompassing krill population biology.