The distorted-wave Born approximation model is recognised as state-of-the-art in predicting acoustic target strength (TS) from fluid-like marine organisms. A stochastic version, the so-called SDWBA model, is endorsed by CCAMLR to predict the TS of Antarctic krill (Euphausia superba) for use in the acoustic echo-integration method. In this study, the SDWBA TS pattern and frequency response behaviours at standard survey frequencies and animal lengths are explored via a sensitivity analysis of some important parameters: number of stochastic realisations, the fatness factor, the material contrasts between the organism and the surrounding seawater and the distribution of animal orientations. Due to the high variability of TS to realistic changes in parameters, it is important that the model should be parameterised according to the potential seasonal variability of krill in the area of investigation. In a multifrequency context, the girth–length relationship for the body shape parameterisation can give a high source of uncertainty, and there is an urgent need for development of accurate and standard procedures to measure both the material property contrasts and the orientation at the time and location of the actual survey. Moreover, a number of incongruities within the CCAMLR SDWBA model package are identified and proposed solutions implemented in an improved package.