Increased availability and affordability of boats, searching technologies (e.g. GPS, sonar) and fishing tackle (e.g. electric reels) have resulted in increased efficiency and diversification of the recreational fishery. Highly efficient specialised recreational fisheries have developed for many species such as swordfish, tunas, sharks, and demersal shelf species, which in some cases, has led to conflict with the commercial sector. As a result, there is a need for reliable estimates of the recreational catch for stock assessment and resource allocation. Unfortunately, obtaining representative data is problematic for specialised recreational fisheries, since they typically lack a complete sampling frame and these fishers are very rare among the wider community. Consequently, traditional sampling methods (e.g. creel or telephone surveys) are inadequate and cost prohibitive for accessing these fishers. Epidemiologists and social scientists routinely survey rare, hidden or hard-to-reach populations within the general community (e.g. HIV carriers, illicit drug users) by penetrating social networks. This poster introduces Respondent-Driven Sampling (RDS), an advanced form of 'snowball' sampling, as a potential cost-effective means of obtaining representative data from elusive specialised recreational fisheries. A project led by the CSIRO and IMAS and funded by the FRDC is currently trialling RDS in the specialised Tasmanian recreational set-line fishery.