Obtaining reliable estimates of important parameters from recreational fisheries is problematic but critical for stock assessment and effective resource management. Sampling methodologies based on traditional design-based sampling theory, is inadequate in obtaining representative catch and effort data, social or demographical characterization, or fisher behaviour from small hard-to-reach components within recreational fisheries (e.g. specialized sport fisheries) that may account for the majority of the catch for some species. A model-based approach to sampling is necessary. Researchers in other disciplines including epidemiology and social sciences routinely survey rare or ‘hidden’ populations within the general community by penetration of social networks rather than by interception of individuals. We encourage fisheries researchers to rethink survey designs and consider the social elements of recreational fishing. Employing chain-referral methods, such as respondent-driven sampling (RDS), may be a statistically robust and cost-effective option for sampling elusive sub-elements within recreational fisheries. Chain-referral sampling methodology is outlined and an example of a complemented ‘RDS-recapture’ survey design is introduced as a cost-effective application to estimating total catch in recreational fisheries.