The alteration of rivers by anthropogenic bank stabilization to prevent the erosion of economically valuable lands and structures has become commonplace. However, such alteration has ambiguous consequences for fish assemblages, especially in large rivers. Because most large, temperate rivers have impoundments, it can be difficult to separate the influences of bank stabilization structures from those of main-stem impoundments, especially because both stabilization structures and impoundments can cause side-channel loss. Few large rivers are free flowing and retain extensive side channels, but the Yellowstone River (our study area) is one such river. We hypothesized that in this river (1) bank stabilization has changed fish assemblage structure by altering habitats, (2) side-channel availability has influenced fish assemblage structure by providing habitat heterogeneity, and (3) the influences of bank stabilization and side channels on fish assemblages were spatially scale dependent. We developed a spatially explicit framework to test these hypotheses. Fish assemblage structure varied with the extent of bank stabilization and the availability of side channels; however, not all assemblage subsets were influenced. Nevertheless, bank stabilization and side channels had different and sometimes opposite influences on the fish assemblage. The effects of side channels on fish were more consistent and widespread than those of bank stabilization; the catches of more fishes were positively correlated with side-channel availability than with the extent of bank stabilization. The influences of bank stabilization and side channels on the relative abundances of fish also varied, depending on species and river bend geomorphology. The variation in river morphology probably contributed to the assemblage differences between stabilized and reference river bends; stabilized alluvial pools were deeper than reference alluvial pools, but the depths of stabilized and reference bluff pools did not differ. The strengths of the relationships among fish assemblages, bank stabilization, and side channels were spatially scale dependent; optimum spatial scales ranged from less than 200 m to 3,200 m up- and downstream, suggesting that bank stabilization and side channels influenced fish assemblages across multiple spatial scales.