In the late 1990s, Idaho’s established stream temperature criteria for the protection of coldwater biota and salmonid spawning were considered inadequate because the criteria did not agree with observed biological conditions in many instances and did not allow for variability in environmental condition or species diversity across a broad area such as the entire State of Idaho. In 2001, benthic invertebrate and fish assemblages in 34 least-disturbed streams in the Salmon River Basin, central Idaho, were evaluated in relation to stream temperature and other environmental variables. The Salmon River Basin retains watersheds that are minimally affected by human activities. These “natural” stream conditions provide a basis for deriving attainable stream temperatures that can be used to set new, and revise existing, water-quality criteria for stream habitats affected by human activities. During July through September 2001, data were collected to document the thermal regime of least-disturbed streams, characterize the distribution of aquatic biota at streams representing a gradient of temperature, and describe the relations between environmental variables and benthic invertebrate and fish assemblages. Nine stream temperature metrics were compared with the U.S. Environmental Protection Agency’s criterion of 10 ° C (degrees Celsius) for bull trout spawning and juvenile rearing. The maximum weekly-maximum temperature at all 33 sites where temperature data were available exceeded this criterion. The maximum daily-maximum temperature (MDMT) at 30 of the sites exceeded the Idaho Department of Environmental Quality (IDEQ) criterion of 13.0 ° C for the protection of salmonid spawning; and the maximum daily-average temperature at all 33 sites exceeded the 9.0 ° C criterion for the protection of salmonid spawning. The thermal regime at most sites did not exceed the IDEQ criteria for the protection of coldwater biota. Nine environmental variables—water-surface gradient, discharge, wetted channel width, width:depth ratio, aspect, total seasonal thermal input, open canopy, riparian canopy density, and elevation were selected for correlation with the nine stream temperature metrics. Elevation showed the strongest inverse correlation with the stream temperature metrics. Two hundred and one benthic invertebrate taxa from the 34 sampling sites were identified. The most abundant taxa were Baetis tricaudatus, Oligochaeta, Tvetenia bavarica gr., Acari, Rhithrogena, Cinygmula, Heterlimnius, Micropsectra, Eukiefferiella devonica gr., Drunella doddsi, and Cricotopus. Of the 201 benthic invertebrate taxa collected during this study, 57 taxa (present at a minimum of 5 sampling sites) were significantly correlated with one or more of the stream temperature metrics. Among the invertebrate taxa collected, 32 were identified as coldwater taxa. Of the coldwater taxa collected, Zapada oregonensis gr. showed the strongest inverse correlation with the stream temperature metrics and was collected at sites where maximum weekly-maximum temperature (based on date of sample and 6 days prior) ranged from 11.3 ° to 18.5 °C. Ten species of fish in the families Salmonidae, Cottidae, and Cyprinidae were collected. Two species (bull trout and chinook salmon) listed as threatened under the U.S. Fish and Wildlife Service Endangered Species Act were collected. Among all species, bull trout showed the strongest inverse correlation between relative fish abundance and stream temperature. Bull trout and juvenile bull trout densities were inversely correlated with stream temperature. The probability of occurrence of juvenile bull trout was significantly correlated with MDMT on the basis of results from a logistic regression model developed during this study. However, this model differed from a similar model developed by the U.S. Forest Service on the basis of regional data collected in the Pacific Northwest. The regression model based on data collected during this study showed higher probabilities of occurrence of bull trout at colder stream temperatures (10 ° to 15 ° C) and lower probabilities of occurrence at warmer stream temperatures (16 ° to 21 ° C) than did the model based on regional data. The model comparisons suggest that regional or local differences need to be considered when deriving stream temperature criteria.