tocks of Patagonian toothfish (Dissostichus eleginoides) in different sectors of the Southern Ocean are considered to be genetically distinct. However, in the Indian Ocean, it is largely unknown whether stocks are separate between shelves and banks separated by deep water. More particularly, the separation of  stocks on the Kerguelen Plateau has not been investigated. This paper examines the assumptions of stock separation using tagged fish from the fishery around Heard Island and McDonald Islands (HIMI) in the Australian Exclusive Economic Zone (EEZ). Movement, growth and trends in fish abundance are examined using these data. The bulk of the data show that fish disperse only a very short distance, up to 15 n miles in most cases, from their point of release. This may be explained in part because of the concentration of fishing in the established grounds, which makes recaptures more likely in the same place as marking. The concentration of effort should also result in a high probability of detecting fish movements between grounds, if they occur, but so far no such movement has been detected. It appears that fish tend to be locally resident during their phase in the depth range of the fishing grounds, but they move on once they approach maturity at about 850 mm total length and become unavailable to the fishery. Four fish have been recorded as moving a long distance, and thus provide contradictory evidence. Three fish moved from ground B of HIMI to Crozet and were all within the size range normally found in the Heard Island fishing grounds, and so appear to be behaving differently from those fish that follow the normal pattern of residency in the grounds. One fish has been recorded as moving from ground A to Kerguelen and was at the upper end of the size range normally found on the grounds. These four long-range movers provide the first documented direct evidence of toothfish moving such distances and of fish moving from one fishery to another. The significance of this is discussed. Trends in abundance of fish show that the number of fish in ground B and in the population as a whole has been reduced since 1998. The implications of these results are discussed. The overall results show that mark–recapture experiments provide important information for the management of toothfish stocks and that trends in abundance derived from these experiments might be useful adjuncts to the assessment process in the future.