Fish biodiversity in rivers is threatened by continual habitat loss. Evidence to support management and restoration of rivers requires information about the specific habitat requirements for entire fish communities. The complexity of this topic, combined with additional challenges of multiple stressors, data limitation, and natural dynamics of rivers, facilitates novel model development for both applied and theoretical arenas. Central to this area of study is the importance of spatial scale. Spatial scale is the context on which measurements, models, management, and policy have the potential to synchronize in an effective manner. The forefront and progress of river restoration depends on the provisioning of timely evidence for decision-making that comes directly from models. This thesis explores and develops a spatial scale approach to model fish communities and habitat in impacted rivers. I present a comprehensive vision of how the spatial scale approach can be implemented across spatial scales. I then introduce new models at each individual scale: micro-scale, meso-scale, macro-scale, riverscape-scale, and beyond-riverscape-scale. For each scale, I studied a different river or system of rivers from around the world. The variety of modeling techniques I used highlighted new opportunities to extend models for methods in river restoration such as flow regime management, fish passage construction, dam removal, non-native fish regulation, and plastic pollution reduction. Advancements for future models and limitations are discussed in the context of spatial scale and interdisciplinarity.