The setting of numeric instream objectives (effects-based criteria) and catchment limits for major agricultural stressors, such as nutrients and fine sediment, is a promising policy instrument to prevent or reduce degradation of stream ecosystem health. We explored the suitability of assemblage thresholds, defined as a point at which a small increase in a stressor will result in a disproportionally large change in assemblage structure relative to other points across the stressor gradient, to inform instream nutrient and sediment objectives. Identification and comparison of thresholds for macroinvertebrate, periphyton, and bacterial assemblages aimed at making the setting of objectives more robust and may further provide a better understanding of the underlying mechanisms of nutrient and fine sediment effects. Gradient forest, a novel approach to assemblage threshold identification based on regression-tree-based random forest models for individual taxa, allowed inclusion of multiple predictors to strengthen the evidence of cause and effect between stressors and multispecies responses. The most prominent macroinvertebrate and periphyton assemblage threshold across the nitrogen (N) gradient was located at very low levels and mainly attributed to declines of multiple taxa. This provided strong evidence for stream assemblages being significantly affected when N concentrations exceed reference conditions and for effects cascading through the ecosystem. The most prominent macroinvertebrate assemblage threshold across a gradient of suspended fine sediment was also located at very low levels and attributed to declines of multiple taxa. However, this threshold did not correspond with periphyton assemblage thresholds, suggesting that the sensitivity of macroinvertebrate assemblages is unrelated to sediment effects on periphyton assemblages. Overall, the spectrum of N concentrations and fine sediment levels within which these stream assemblages changed most dramatically were relatively narrow given the wide gradients tested. We conclude that assemblage thresholds can inform the setting of generic instream nutrient and sediment objectives for stream ecosystem health. For example, the most stringent objective for instream N concentration should be set at values similar to reference concentrations for full protection of sensitive taxa or overall stream biodiversity. To avoid severe degradation of stream biodiversity, the least stringent N objective should stay well below the point where significant turnover subsided.