Reunion Island is in the South‐West Indian Ocean (SWIO), where all freshwater fish species are diadromous. The ecological status assessments of freshwater in watersheds have revealed a continuing deterioration in these fish populations due to anthropic pressures. In this context, monitoring the fish's biological sustainability is crucial to ensure the health of these estuarine ecosystems. The aim of this study was to compare the efficacy of conventional electrofishing monitoring (EF) with the environmental DNA metabarcoding tool to evaluate fish biodiversity in the estuaries. We measured the diversity and structure of the fish community in three estuaries with various geographical, hydrological, and anthropogenic conditions over different seasons. To this end, fish were captured by EF, and we then isolated DNA from the water samples to perform bioinformatic analyses derived from eDNA, using the 12S marker. Statistical analyses were carried out to compare the results of these two methods. For all watersheds combined, a comparison of the results for measuring fish richness showed that eDNA performed significantly better than EF. Indeed, the eDNA detected 31 species, whereas the EF detected only 12 species. For both methods, we observed significant differences in community structure between watersheds, with a significant nestedness phenomenon where the fish assemblage obtained from EF captures is a sub‐assemblage of that obtained from eDNA. Moreover, compared to EF, eDNA enabled the detection of endemic to the Mascarene region species (e.g., Cotylopus acutipinnis), introduced exotic species (e.g., Oreochromis niloticus), and species difficult to capture and identify due to their juvenile life stage through EF (e.g., Anguilla sp.). Our data confirm the effectiveness of eDNA to detect fish species, both taxonomically and in terms of species richness and proves to be an effective tool for monitoring fish diversity of the islands of the SWIO.