1. To manage the environmental flow requirements of sedentary taxa, such as mussels and aquatic insects with fixed retreats, we need a measure of habitat availability over a variety of flows (i.e. a measure of persistent habitat). Habitat suitability measures in current environmental flow assessments are measured on a ‘flow by flow’ basis and thus are not appropriate for these taxa. Here, we present a novel measure of persistent habitat suitability for the dwarf wedgemussel (Alasmidonta heterodon), listed as federally endangered in the U.S.A., in three reaches of the Delaware River. 2. We used a two-dimensional hydrodynamic model to quantify suitable habitat over a range of flows based on modelled depth, velocity, Froude number, shear velocity and shear stress at three scales (individual mussel, mussel bed and reach). Baseline potentially persistent habitat was quantified as the sum of pixels that met all thresholds identified for these variables for flows ≥40 m3 s−1, and we calculated the loss of persistently suitable habitat by sequentially summing suitable habitat estimates at lower flows. We estimated the proportion of mussel beds exposed at each flow and the amount of change in the size of the mussel bed for one reach. 3. For two reaches, mussel beds occupied areas with lower velocity, shear velocity, shear stress and Froude number than the reach average at all flows. In the third reach, this was true only at higher flows. Together, these results indicate that beds were possible refuge areas from the effects of these hydrological parameters. Two reaches showed an increase in the amount of exposed mussel beds with decreasing flow. 4. Baseline potentially persistent habitat was less than half the areal extent of potentially suitable habitat, and it decreased with decreasing flow. Actually identified beds and modelled persistent habitat showed good spatial overlap, but identified beds occupied only a portion of the total modelled persistent habitat, indicating either that additional suitable habitat is available or the need to improve habitat criteria. At one site, persistent beds (beds where mussels were routinely collected) were located at sites with stable substratum, whereas marginal beds (beds where mussels were infrequently collected or that were lost following a large flood event) were located in scoured areas. 5. Taken together, these model results support a multifaceted approach, which incorporates the effects of low and high flow stressors, to quantify habitat suitability for mussels and other sedentary taxa. Models of persistent habitat can provide a more holistic environmental flow assessment of rivers.