Glacier-fed stream fauna may be severely altered by climate change due to the melting of glaciers, therefore, there is an urgent need to monitor glacier-fed stream diversity. Environmental DNA (eDNA) is a very promising method that could allow for large scale cost-effective biomonitoring. However, no literature has been published detailing the use of eDNA metabarcoding on high-altitude alpine stream fauna. In this study, we performed an eDNA survey and a traditional morphological survey on the Avançon de Nant stream, a small glacier-fed stream in the Vaud Canton in Switzerland. We aimed to 1) test the validity of eDNA metabarcoding as an effective tool for biodiversity assessments, 2) compare eDNA and traditional survey methods, 3) benchmark crucial steps for eDNA routine surveys and 4) perform alpha and beta diversity measurements using eDNA on the animal community of the Avançon de Nant stream and its watershed. For this, we designed and tested primer sets that would discriminate against fungi stricto sensu, one of the main sources of untargeted eDNA, while still amplifying a broad spectrum of targeted organisms. These primers were then used in our eDNA survey and successfully excluded fungi species. For our morphological survey, we collected individuals using a kicknet every two and a half months, which were identified to the genus-level. 3x1L of water were also sampled for our eDNA survey. Water was filtered on a 0.22µm filter. eDNA was extracted, amplified using the most effective primer set we designed and sequenced. We found that eDNA not only detected more macroinvertebrate taxa than our morphological survey, but it also detected terrestrial fauna from the overall watershed, thus providing additional information. It also allowed for lower levels of identification (i.e. species assignation) and alpha or beta diversity measurements. When considering only aquatic-taxa, richness variations across dates or between seasons were non-significant. No clear community composition turnovers were encountered, suggesting that dates and seasons had little impact on the results of eDNA survey of benthic macroinvertebrate. Conversely, terrestrial taxa richness differed significantly between dates and turnovers in community compositions were detected. Therefore, depending on the targeted taxa, eDNA can either be sampled regardless of the season or should be sampled multiple times throughout the year to avoid underestimation of biodiversity, at least in the Avançon de Nant stream and within its watershed. Nonetheless, not all species captured in our morphological survey were detected in our eDNA survey. Moreover, only a fraction of our sequences was assigned (∼14.89%), which is similar to the results obtained by other studies focusing on invertebrate communities characterization using eDNA. Morphological and molecular data were sometimes incongruent, possibly due to an insufficient number of sampling dates in our morphological survey. Species accumulation curves revealed incomplete sampling, with 20% to 50% of missing diversity depending on the estimation methods. This highlights actual limitations of eDNA metabarcoding, such as the use of degenerated primers, which guarantee taxonomic breadth but increase the proportion of untargeted amplification. It may be more suitable to decrease taxonomic breadth in preference for additional selectivity. Restricting the number of taxa could allow for the design of more selective primers, which could help reduce the proportion of untargeted amplification. This could decrease the cost of multiple replicates and sequencing depth.