River ecosystems are highly dynamic systems, governed by the temporal variation of their flows. Riverine fish have evolved different life histories in response to these flow regimes, represented by temporal sequences of floods and droughts. Although there is a growing need to forecast how riverine fish will respond to rapidly changing flow regimes, data challenges often preclude robust models for many species. Flow response guilds, which represent groups of species that respond similarly to flow regimes and tend to share similar traits, have the potential to serve as modelling units for such models, thereby increasing the transferability of models among systems where species-specific information is lacking. Using New Zealand as a case study, we developed flow response guilds. We used a series of flow metrics that characterize different components of the flow regime, and fish occurrence data to build hierarchical models that analyse how individual species respond to the different flow faucets: the ‘flow niche’. We then clustered these individual flow niches into fish-flow response guilds that represent units that can be later modelled in predictive flow-driven population or community models. These fish-flow response guilds can be a useful first step to develop mechanistic flow-population models for systems that are lacking species-specific information. By understanding this fish-flow relationship, more system-specific conservation strategies could be designed by targeting the flow conditions necessary to preserve rivers biodiversity.