Integrated models of recreational fisheries as social–ecological systems contain—at a minimum—two dynamically linked components: a fishing effort or harvest dynamics sub-model (representing recreational fisher behaviour) and a population dynamics sub-model (representing fish dynamics in response to exploitation). Here, we review and categorize the use of integrated models, provide a set of general instructions for building them, and identify gaps and opportunities for further development. The structure of coupled social–ecological models diverges along two major paths: agent-based models that follow the behaviour of individual autonomous agents (generally fishers or fish) and models that track the aggregate dynamics of a fish population and the fishing effort exerted on it by fishers. Most integrated models published so far are lopsided in their development. That is, they often contain one sub-model (harvest or population dynamics) that is detailed and grounded in empirical data, whereas the other sub-model is a more generic representation of that process. The future of integrated models depends on increased collaboration between the social and ecological sciences, a rigorous quantification of both fish and fisher behavioural patterns, and a confrontation of model predictions with actual system behaviour.