Fish assemblages and environmental variables were evaluated from 37 least-disturbed, 1st-through 6th-order streams and springs in the upper Snake River basin, western USA. Data were collected as part of the efforts by the U.S. Geological Survey National Water Quality Assessment Program and the Idaho State University Stream Ecology Center to characterize aquatic biota and associated habitats in least-disturbed coldwater streams. Geographically, the basin comprises four ecoregions. Environmental variables constituting various spatial scales, from watershed characteristics to instream habitat measures, were used to examine distribution patterns in fish assemblages. Nineteen fish species in the families Salmonidae, Cottidae, Cyprinidae, and Catostomidae were collected. Multivariate analyses showed high overlap in stream fish assemblages among the ecoregions. Major environmental factors determining species distributions in the basin were stream gradient, watershed size, conductivity, and percentage of the watershed covered by forest. Lowland streams (below 1,600 m in elevation), located mostly in the Snake River Basin/High Desert ecoregion, displayed different fish assemblages than upland streams (above 2,000 m elevation) in the Northern Rockies, Middle Rockies, and Northern Basin and Range ecoregions. For example, cottids were not found in streams above 2,000 m in elevation. In addition, distinct fish assemblages were found in tributaries upstream and downstream from the large waterfall, Shoshone Falls, on the Snake River. Fish metrics explaining most of the variation among sites included the total number of species, number of native species, number of salmonid species, percent introduced species, percent cottids, and percent salmonids. Springs also exhibited different habitat conditions and fish assemblages than streams. The data suggest that the evolutionary consequences of geographic features and fish species introductions transcend the importance of ecoregion boundaries on fish distributions in the upper Snake River basin.Many rivers and streams in the conterminous terns (Karr 1991). However, before the effects of United States have been degraded as a result of human alterations of streams can be evaluated, binonpoint-source pollutants, fragmentation (i.e., ological criteria for water quality monitoring redams and diversions), habitat alteration, and in-quire data based on least-disturbed or reference troduction of nonnative fish species (Moyle 1986; streams or other suitable historical data (Hughes Heede and Rinne 1990;Allan and Flecker 1993; e t al. 1986;Hayslip 1993). For example, appli-Doppelt et al. 1993;Dynesius and Nilsson 1994). cation of the index of biotic integrity (IBI) depends Human alterations of the physical, chemical, or on reg ional reference site information to score inbiological properties of lode systems usually result dividual fish metrics (Karr 1991) The asses sment in changes in the distribution and structure of fish of fish assemblages in re i at ion to least-disturbed assemblages^ Docume...