Environmental DNA (eDNA) is a powerful tool for biological monitoring with the potential to overcome weaknesses of conventional macroinvertebrate surveys in running waters. However, the ability of eDNA to detect changes of macroinvertebrate communities immediately downstream of a perturbation, particularly in upstream to downstream comparisons, has not been adequately explored. To address this issue, we investigated the potential of eDNA based assessment of a perturbation by comparing the results with macroinvertebrate surveys in a river influenced by the inflow of a metal contaminated tributary. Both methods revealed distinctly lower richness of taxa and zero radius operational taxonomic units (ZOTUs) at the metal contaminated tributary site compared to other study sites. Results from collection of macroinvertebrates indicated that most richness and abundance metrics were significantly reduced at three metal contaminated sites located 150 to 750 m downstream from the inflow of the tributary. In contrast, the eDNA based assessment revealed similar ZOTU richness at a reference site and the downstream, contaminated sites. Although statistically not significant because sample sizes were small, eDNA based nonmetric multidimensional scaling revealed some separation between the reference site and two downstream sites. However, no separation was apparent between the reference site and the site immediately downstream. This result suggested that eDNA at a site 150 m downstream from the inflow was likely affected by downstream drift of eDNA from the upstream reference area. That drift complicated assessment of the community a short distance from the perturbation. The site separation detected by eDNA based assessment was promising, but the ZOTUs that contributed to the separation were mainly from dipteran taxa rather than from metal sensitive mayflies, which were significantly lower in abundance at the downstream, contaminated sites. Developing reliable local DNA barcoding information, particularly for these mayflies, may help overcome the limitations of making evaluations over relatively small spatial scales, such as upstream-downstream comparisons.