This research tested the ability of fish morphology to predict membership of fishes in habitat guilds, their swimming performance, and habitat preference. Further, it considered methods for choosing a surrogate species to identify habitat of target species. Morphological discriminant functions were developed using morphological traits of fishes from one river to identify membership in two habitat guild systems (mesohabitat and microhabitat). Functions were then used to test factors influencing classification success of holdout tests and validated using fishes of a second river. Morphology was only partly successful (50%) at predicting membership in habitat guilds. Morphology identified species by shape, i.e., classifying test species into guilds with members of their genus, but not habitat use, because morphology and habitat were not strongly linked through function. By improving guild definition, relationships between morphology and habitat (Froude number) were identified for all fish groups examined (darters, benthic minnows, pelagic minnows, and suckers). Relationships were not transferable among groups. Further, morphology of eight minnows was linked to swimming performance, a key task for using habitat, in lab measurements of critical swimming speeds. In turn, swimming performance was related to habitat (Froude number). Morphology will be most successful at predicting habitat use of fishes when (1) more, discrete guilds are used, (2) guilds are identified within families, (3) variation in lifestyles (benthic vs. pelagic) is considered, and (4) key tasks related to using habitat are strongly associated with morphology. Finally, I examined a phylogenetic approach to identifying useable habitat. Closely related surrogate species were not more accurate in identifying habitat of target species than surrogates chosen by other methods. When a target species used only one mesohabitat, the highest overlap in habitat use occurred with other fishes of the same family using that mesohabitat (within a physiographic province). For target species using several mesohabitat types, surrogates from the next highest taxonomic unit, e.g., genus or subgenus, provided the most accurate information. Ecomorphology offers a mechanistic and defensible method for identifying habitat preferences of fishes and should be more widely considered as a tool for establishing habitat relationships of stream fishes.