Chrosomus cumberlandensis (Blackside Dace) is a small-bodied, freshwater fish endemic to the upper Cumberland River system in southeastern Kentucky and northeastern Tennessee. A detailed study of its habitat requirements using presence-absence data has not been published to date. Identification of important habitat variables at multiple spatial scales would facilitate proactive management and recovery of this federally listed species. Using logistic regression, we developed and validated habitat models of Blackside Dace presence as a function of environmental variables measured at two spatial scales (i.e., stream and reach). For model development, map-produced and field-collected variables were gathered for 91 waterways at the stream scale and 72 localities at the 200-m reach scale during summer. Our initial models predicted the likelihood of Blackside Dace presence to be optimized in streams with crude gradient between 1% and 6%, and in reaches with a turbidity ≤10 NTU, dissolved oxygen >8.5 mg/L, summer water temperatures between 14.6 °C and 18.5 °C, conductivity <240 µS, percent riffle between 35% and 50%, and link magnitude between 3 and 6, although the species was occasionally observed in locations with conditions outside of this predicted optimal range. We then validated the models by collecting additional data from 27 new streams and 47 new reaches. Model performance was assessed with Cohen's kappa (κ;). The strongest models included conductivity as a predictor variable, with the combination of conductivity and temperature producing the strongest performance (κ; = 0.41). Models containing crude gradient, turbidity, dissolved oxygen, and percent riffle generally did not perform well upon validation. Our findings suggest that conductivity, water temperature, and link magnitude are three important reach-scale variables for resource managers to consider when conserving populations of Blackside Dace.