Landscape genetics is a field dealing with local genetic differences and contributes to strategic conservation planning. Recently, environmental DNA (eDNA) metabarcoding has proven useful not only for detecting species but also for assessing genetic diversity and genetic structure on a large scale such as in phylogeography. However, it remains unclear whether eDNA analysis also has sufficient power to perform the landscape genetics, which focuses on a local scale. To reveal the applicability of eDNA to landscape genetics, we conducted an eDNA metabarcoding analysis of the mitochondrial DNA D-loop region of the fluvial sculpin Cottus nozawae in the upper Sorachi River in Japan and compared the results with inferences based on traditional tissue-based approaches. As a result, the spatial distribution of haplotypes was generally consistent between the eDNA- and tissue-based approaches. In addition, the genetic differentiation statistics calculated using eDNA and tissue samples were highly correlated when compared in the same genetic region. Furthermore, if an inference based on genome-wide SNP data from tissue samples was taken as known truth, the inference by eDNA was not necessarily inferior to the inference by the same region from tissue samples. Finally, we confirmed that analyses using eDNA data can reveal patterns such as isolation-by-distance shown in previous studies on this species, indicating the applicability of eDNA to basic landscape genetics. Even though some limitations remain, eDNA may have great potential for conducting basic landscape genetics.