Estuarine ecosystems are threatened by numerous anthropogenic pressures. Fish assemblages are a dominant component of estuarine macrofauna and serve as indicators for the health of these transitional water ecosystems. Environmental DNA (eDNA) metabarcoding is increasingly used to assess the biodiversity of fishes in estuaries. However, there is a need to further establish how effective eDNA metabarcoding can be relative to conventional fish sampling methods across multiple estuaries and seasons. This study compared fish assemblages detected via eDNA metabarcoding of surface water samples to contemporary sampling with conventional fishing gears in three temperate estuaries (UK), during early summer and autumn. Most species caught by fishing were detected by eDNA. Species richness estimates from eDNA were two to ten times higher than estimates based on fishing, and included taxa of conservation importance and a non-native species. The eDNA assemblage composition was significantly different to the assemblage detected by seine nets. Importantly, eDNA methods could effectively discriminate between fish assemblages of different estuaries and seasons. Fish assemblages in estuaries are often not monitored due to resource constraints. The dynamic nature of estuaries may make fishing gear deployment difficult and inconsistent. The findings indicate that eDNA metabarcoding is suited to gathering large amounts of information on fish biodiversity, at a relatively low sampling effort, compared to established fishing methods. Therefore, eDNA shows promise as an assessment tool for fish assemblage structure and ecosystem health in estuarine environments, with application to statutory monitoring.