A krill density model suggested during WG-EMM at Kochi ( Dy?(1-Rl)-Dy-1?e-M=0) still involved uncertainties of age 1 krill. Present document improves the model by taking the uncertainties in to account, and assuming a reasonable mortality. The model suggested that if we expect the currently accepted values of mortality M (0.8-1.0), the potential proportional recruitment should be larger than the observed values. Although potential recruitments were incorporated, and also reasonable mortality was used in the model, dramatic variation of densities after 1994/95 season could not be clearly explained.