• Gradient forest (GF) models evaluate taxon specific responses to stressors. • GF diatom, invertebrate, fish models showed several consistent variables selected. • Variables included substrate size, dissolved oxygen, and various contaminant measures. • Nitrogen and phosphorus important in both diatom and invertebrate models but not fish. • Taxa responded to multiple stressors following patterns of sensitive versus tolerance. This study is part of the regional stream-quality assessment (RSQA) conducted by the U.S. Geological Survey (USGS) National Water Quality Assessment (NAWQA) project. The purpose of this study is to examine small streams along land-use and stressor gradients at the regional scale and to evaluate the relative importance of instream stressors on diatom, macroinvertebrate, and fish assemblages. In 2015, the RSQA project assessed stream quality in 82 wadeable streams that were selected along an urban land-use gradient in the Pacific Northwest Region (PNW) of the United States. This study evaluates the effects of four major categories of measured instream stressors – flow (i.e. alteration), water quality, habitat, and contaminants (in water and sediment) – on stream biota. We used gradient forest (GF) models to evaluate taxon specific responses to the various stressors for the three biotic assemblages. Results for diatom, invertebrate and fish assemblages showed that several environmental variables including substrate size, dissolved oxygen, and two or more different contaminants were selected in each of the GF models. In general, all three assemblages were negatively associated with any contaminant measures above zero, except the more tolerant taxa in each assemblage, which responded positively to contaminants. Total nitrogen (TN) and total phosphorus (TP) were important in both the diatom and invertebrate GF models but not in the fish models, which were related to temperature and stream flow. TP and TN were the top two variables for diatom GF models and various taxa responded at a range of nutrient concentrations; however, some taxa responded at low concentrations, for example around 0.02 for TP and 0.5 mg/L for TN. In general, the three biotic assemblages responded to multiple stressors following general patterns of known sensitive versus tolerant taxa for each of the biotic groups studied, yet the GF models allow us to explore taxon specific responses. For example, most of the sensitive Ephemeroptera, Plecoptera, Trichoptera invertebrate taxa (EPT) responded negatively when any contaminant increased above zero; yet some taxa such as the tolerant Trichoptera Cheumatopsyche responded positively to contaminants and many of the other stressors. The findings of this study demonstrate the value of using multiple assemblages to monitoring stressor gradients associated with urban stream systems and the importance of evaluating the responses of individual taxa to stressors.