Adoption of the ecosystem approach to fisheries management relies on recognition of the link between fish and other components of the ecosystem, namely their physical and biological habitat. However, identifying the habitat requirements of marine fishes and hence determining their distribution in space and time is scientifically complex. We analysed the methodologies and findings of research on temperate, demersal fish habitat requirements to highlight the main developments in this field and to identify potential shortfalls. Many studies were undertaken over large spatial scales (≥100s km2) and these generally correlated abundances of fish to abiotic variables. Biological variables were accounted for less often. Small spatial scale (≤m2), experimental studies were comparatively sparse and commonly focused on biotic variables. Whilst the number of studies focusing on abiotic variables increased with increasing spatial scale, the proportion of studies finding significant relationships between habitat and fish distribution remained constant. This mismatch indicates there is no justification for the tendency to analyse abiotic habitat variables at large spatial scales. Innovative modelling techniques and habitat mapping technologies are developing rapidly, providing new insights at the larger spatial scales. However, there is a clear need for a reduction in study scale, or increase in resolution additional to the integration of biotic variables. We argue that development of sound predictive science in the field of demersal fish habitat determination is reliant on a change in focus along these lines. This is especially important if spatial management strategies, such as Marine Protected Areas (MPA) or No Take Zones (NTZ), are to be used in future ecosystem-based approaches to fisheries management.