This paper presents an assessment of the stock of toothfish around South Georgia (Subarea 48.3) using the CASAL stock assessment software (Bull et al., 2005). Detailed attention is given to the incorporation of as much of the available tuning data as possible, as well as a whole range of assessment sensitivities – to fixed parametric assumptions, model structures and alternative data scenarios. Given the integrated nature of the assessment, particular attention is given to rigorous statistical weighting of the various tuning datasets. Bayesian methods are used in the estimation procedure, and uncertainty in the dynamics is explored using Markov Chain Monte Carlo (MCMC) methods; methods for fast approximations to the more time-consuming MCMC tools and CASAL-specific convergence checking tools are also detailed. Finally, long-term yield calculations were undertaken, given the CCAMLR decision rules, for five main assessment candidates.