Insufficient water sample volumes can be a limiting factor for detecting species with environmental DNA (eDNA) from aquatic habitats. We compared detections of freshwater mussel (Unionidae) communities using large water sample volumes and dead‐end hollow fiber ultrafiltration (D‐HFUF or DEUF) with traditional eDNA filtration methods that use relatively small water sample volumes. Unionid species were detected in approximately 50‐L D‐HFUF eDNA samples with two mitochondrial DNA metabarcoding markers (COI and ND1) and compared to species detection results from eDNA captured from commonly used 1‐L samples filtered with polyethersulfone (PES) filters at three lotic sites in Georgia and Missouri. Of the 431,560 COI and 1,035,472 ND1 reads from all environmental samples of both filter types that passed quality control, 95% (410,755 reads) of COI reads and 85% (883,472 reads) of ND1 reads were assigned to a unionid species. Nineteen different freshwater mussel species were detected across all D‐HFUF samples, and 11 species were detected across all PES samples. Reads assigned to the genus Elliptio could not be resolved beyond the genus level with either marker. From D‐HFUF samples, 15 and 16 mussel species were detected with the COI and ND1 markers, respectively. From PES samples, nine and seven species were detected with the COI and ND1 markers, respectively. More mussel species were detected at each site in D‐HFUF samples than in PES samples regardless of whether results from both markers were combined or evaluated separately. Our results demonstrate the merit of further exploration and optimization of D‐HFUF for capturing eDNA from high‐volume water samples to facilitate detection of unionids and likely other aquatic organisms.