Data collected on board the FV American Champion during Phases 1 and 2 of the experimental crab fishery were analysed using generalised additive models and depletion estimators. Results from the generalised additive models show that the density of fishable Paralomis spinosissima is highest off the northern coast of South Georgia and at depths of about 100 to 300 fathoms. The Phase 1 results suggest that it would not be appropriate to extrapolate local estimates of abundance to the whole of Subarea 48.3 solely on the basis of depth-specific seabed area; extrapolations must consider location. Linear models fitted to catch per unit effort (CPUE) and cumulative catch data from the Phase 2 depletion experiments did not have significant negative slopes. The insignificant regressions were probably a result of small catches, inter-haul variability in CPUE, and crab movement and suggested that depletion estimators will not be appropriate tools for estimating local abundances of P. spinosissima. It was not possible to determine whether depletion estimators can be used to estimate crab abundance when larger areas are considered and larger catches are taken. Approximately 6 000 crabs were tagged and released during Phase 2. Four of the tagged crabs were recovered, and these crabs were at liberty for about one to five weeks. The recaptured crabs had minimum movement rates of 0.08 to 0.25 n miles/day. Mark-recapture estimates of standing stock and density were made for the area around Phase 2's third depletion square. The assumptions of the mark-recapture model were probably violated, but attempts were made to account for biased sampling of recaptured crabs and the movement of crabs away from their release sites. Density estimates from the mark-recapture data were in the order of 50 000 to 100 000 legal-sized male crabs/n mile², and there was a lot of uncertainty associated with these estimates. In general, the experimental harvest regime was successful. The regime provided important information about crab distribution and facilitated evaluation of local depletion estimators for use in stock assessment, but the CCAMLR Working Group on Fish Stock Assessment (WG-FSA) may wish to re-evaluate the design of Phase 2 and consider the implementation of a wide-scale, intensive tagging study.