Diverse communities of groundwater-dwelling organisms (i.e., stygobionts) are important for human wellbeing. However, we lack an understanding of the factors driving their distributions, making it difficult to protect these at-risk species. Obtaining distribution data for stygobionts is diffiult because subterranean environments are some of the most difficult to sample. New sampling methods, such as environmental DNA (eDNA), show promise to improve stygobiont detection, but sources of sampling bias are poorly understood. Our study objectives were to quatify detection probability using traditional visual surveys and eDNA surveys for both cavefishes and cave crayfishes and determine the landscape factors related to the occurrence of cavefishes and cave crayfishes in the Ozark Highlands ecoregion, USA. We sampled cavefishes and cave crayfishes at 61 sampling units using both visual and environmental DNA surveys. We then modeled occurrence probability in relation to lithology and human disturbance while accounting for imperfect detection. The effectiveness of eDNA or visual surveys varied by environmental conditions (i.e., water volume, prevailing substrate, and water velocity) and the target taxa. Our results also indicated that occurrence probability of cave crayfishes was negatively associated with human disturbance, whereas there was a weak positive relationship between cavefish occurrence and disturbance. Both cavefishes and cave crayfishes were more likely to occur in limestone rather than dolostone lithology. Individual cave crayfish species were restricted to dolostone or limestone lithology; however, the Ozark cavefish (Troglichthys rosae) was detected in both lithology types. Our results indicate structuring factors are related to the distribution of these taxa, but with human disturbance as a prevalent modifier of distributions for crayfish taxa. Limiting human alteration near karst features may be warranted to promote the persistence of some stygobionts. Moreover, our results suggest current sampling efforts are inadequate to detect cryptic species; therefore, expanding sampling may be needed to develop effective conservation actions.