Monitoring aquatic biodiversity through DNA extracted from environmental samples (eDNA) combined with high-throughput sequencing, commonly referred to as eDNA metabarcoding, is increasing in popularity within the scientific community. However, sampling strategies, laboratory protocols and analytical pipelines can influence the results of eDNA metabarcoding surveys. While the impact of laboratory protocols and analytical pipelines have been extensively studied, the importance of sampling strategies on eDNA metabarcoding surveys has not received the same attention. To avoid underestimating local biodiversity, adequate sampling strategies (i.e. sampling intensity and spatial sampling replication) need to be implemented. This study evaluated the impact of sampling strategies along an altitudinal and biodiversity gradient in the upper section of the Murrumbidgee River (Murray-Darling Basin, Australia). An eDNA metabarcoding survey was used to determine the local fish biodiversity and evaluate the influence of sampling intensity and spatial sampling replication on the biodiversity estimates. The results show that optimal eDNA sampling strategies varied between sites and indicate that river morphology, species richness and species abundance affect the optimal sampling intensity and spatial sampling replication needed to accurately assess the fish biodiversity. While the generality of the patterns will need to be confirmed through future studies, these findings provide a basis to guide future eDNA metabarcoding surveys in river systems.