Fish passes, which are designed to promote the free passage of fish past riverine obstructions, generally perform poorly for the entire community and even target species are not able to pass as well as previously thought. This is often because: 1) Fundamental knowledge of how fish interact with the complex hydrodynamic conditions within passes is lacking, 2) passage technology is less well developed for weaker swimming non-salmonid species, and 3) fish display complex behaviours, such as rejecting accelerating velocity gradients associated with downstream bypass intakes. This thesis addresses these issues. Current understanding on how fish interact with complex flows is discussed, and limitations and knowledge gaps highlighted. Previous studies in this field have generally focussed on identifying correlative links between one of any number of hydrodynamic metrics. However, often the causal reason behind these links is obscure. This issue was addressed by returning to first principles and experimentally investigating the behaviour of brown trout, Salmo trutta, under the simple assumption that space use should be governed by energy conservation strategies. The results indicate that fish use space as predicted; through either the selection of low drag regions or where they could express specialised energy reducing behaviours (e.g. the Kármán gait). A simple, robust and biologically relevant hydrodynamic descriptor of drag that can be used as a proxy for the energetic cost of holding station in a turbulent flow is described and tested and two new specialised behaviours identified (wall holding and tail holding). European eel, Anguilla anguilla, and river lamprey, Lampetra fluviatilis, are both weaker swimming non-salmonid fish, which are in decline, and for which conventional fish passes perform poorly. Experimental trials were undertaken to quantify the efficiency of a new method for improving the upstream passage of eel and lamprey at a model crump weir. Side-mounted and vertically oriented bristle passes improved the upstream passage of both species although there was interspecific differences in their efficacy, with the passes being more effective for eel than lamprey. Behavioural observations of both species as they used the bristle passes will aid in optimisation of this and similar pass types. According to Signal Detection Theory (SDT), the ability to detect a signal (discriminability) decreases with increasing levels of internal and/or external noise. Brown trout were used to test whether hydrodynamic noise would mask the detection of an accelerating velocity gradient as fish moved downstream. The experimental results were inconclusive but they represent the first attempt to use SDT as a tool to manipulate animal behaviour and aid in the conservation of vulnerable species. As such they provide a useful platform for future research. Experimental research presented within this thesis has advanced scientific knowledge that will aid in the development of methods to improve fish passage at migration barriers. The changes brought about as a result of this research will help conserve vulnerable fish species, something that should, in turn, help promote productive and resilient ecosystems that benefit society.