Understanding a species’ thermal niche is becoming increasingly important for management and conservation within the context of global climate change, yet there have been surprisingly few efforts to compare assessments of a species’ thermal niche across methods. To address this uncertainty, we evaluated the differences in model performance and interpretations of a species’ thermal niche when using different measures of stream temperature and surrogates for stream temperature. Specifically, we used a logistic regression modeling framework with three different indicators of stream thermal conditions (elevation, air temperature, and stream temperature) referenced to a common set of Brook Trout Salvelinus fontinalis distribution data from the Boise River basin, Idaho. We hypothesized that stream temperature predictions that were contemporaneous with fish distribution data would have stronger predictive performance than composite measures of stream temperature or any surrogates for stream temperature. Across the different indicators of thermal conditions, the highest measure of accuracy was found for the model based on stream temperature predictions that were contemporaneous with fish distribution data (percent correctly classified = 71%). We found considerable differences in inferences across models, with up to 43% disagreement in the amount of stream habitat that was predicted to be suitable. The differences in performance between models support the growing efforts in many areas to develop accurate stream temperature models for investigations of species’ thermal niches.