Assessing aquatic biodiversity in remote waterways using traditional fish sampling methods is expensive and time-consuming. For this reason, many waterways in northern Australia remain poorly studied, despite the fact that they harbour high biodiversity. We aimed to develop metabarcoding methods for fish richness assessment in waterways across tropical Queensland, Australia, using practical field methods for citizen-scientist application. Five replicate water samples plus a field blank were collected and preserved at 12 sites by volunteers from the CreekWatch monitoring program. Two sampling events were carried out at each sampling site, during the wet and dry seasons. We averaged four volunteers and 24 h of volunteer time for each sampling time and location. Two universal assays amplifying the 16S and COI gene regions were used to characterize fish richness. A literature search identified 106 fish species to be likely present in the catchments sampled, including 97 native species and 9 invasive species, from 12 Orders, 34 Families and 67 Genera. Although eDNA metabarcoding data revealed overlap with the species found in the literature search, some common species were not identified. We also found invasive and native species that had not been previously described at some sites, but that their presence is plausible. Two amplicon sequence variants could not be assigned to a particular operational taxonomic unit at one sampling site, highlighting the need to grow reference database to include northern Australia species. Our results highlight the utility of eDNA metabarcoding for fish richness assessment, as well as some shortcoming of the technique.