We conducted several different assessments of toothfish in Subarea 48.3 using four different methods, one fishery dependent and three fishery-independent. Survey-generated recruitment is apparently not sufficient to generate realistic estimates of current biomass using GYM. ASPM runs coded in AD Modelbuilder failed to fit satisfactorily to CPUE trends and catch length composition. The most consistent results were provided by tag-recapture data adjusted for initial tag mortality, natural mortality, tag loss rates and fishery age-based selectivity. Estimations based on local depletions of toothfish generated results essentially similar to the tagging data. Based on these results, we conclude that the most likely current exploitable biomass of toothfish in Subarea 48.3 is between 45,000 and 62,000 t. GYM was run to generate these levels of exploitable biomass at the start of 2004. This generated estimates of yield between 3240 and 4150 tonnes.