Biodiversity surveys are crucial to monitor the health of threatened aquatic ecosystems, such as tropical estuaries and mangroves. Conventional monitoring methods are intrusive, time-consuming, substantially expensive and frequently only provide rough assessments in complex habitats. Recent advanced molecular methods such as environmental DNA (eDNA) using high-throughput sequencing technology are promising although only few applications in tropical estuarine ecosystems have been reported. In this study, we explore the advantages and limitations of an eDNA metabarcoding survey on the fish community of the Merbok Estuary (Peninsular Malaysia). COI and 12S eDNA metabarcoding assays collectively detected 178 species belonging to 127 genera, 68 families, and 25 orders. This approach captured significantly more species than in any previous traditional surveys, including a few species of conservation importance. However, we highlight three limitations: (1) in the absence of a comprehensive reference database the identities of several species are unresolved; (2) a fraction of previously documented specimen-based diversity was not captured by the current method, may be as a consequence of PCR primer specificity, and (3) the detection of non-resident species; stenohaline freshwater taxa (e.g. cyprinids, channids, osphronemids) and marine coral reef taxa (e.g. some sharks, holocentrids and syngnathids), not known to frequent estuaries, leading to the supposition that their DNA have drifted into the estuary through water movement. The community analysis revealed that fish diversity along the estuary is not homogenous, with the upstream more diverse than further downstream. This may be the consequence of the salinity or pollution gradients. In summary, we demonstrated the practicality of eDNA metabarcoding in assessing fish community and structure within a complex and rich tropical environment within a short sampling period. However, some limitations need to be considered and addressed to fully exploit the efficacy of this approach, in particular the development of a comprehensive reference genetic database.