To monitor the effect of nature restoration projects in North Sea ecosystems, accurate and intensive biodiversity assessments are vital. DNA-based techniques and especially environmental (e)DNA metabarcoding is becoming a powerful monitoring tool. However, current approaches rely on genetic target regions under 500 bp, offering limited taxonomic resolution. We developed a method for long-read eDNA metabarcoding, using Nanopore sequencing of a longer amplicon and present DECONA, a read processing pipeline to enable improved identification of marine vertebrate species. We designed a universal primer pair targeting a 2 kb region of fish mitochondrial DNA and compared it to the commonly used MiFish primer pair targeting a ~ 170 bp region. In silico testing showed that 2 kb fragments improved accurate identification of closely related species. Analysing eDNA from a North Sea aquarium showed that sequences from both primer pairs could be assigned to most species, and additional species level assignments could be made through the 2 kb primer pair. Interestingly, this difference was opposite in eDNA from the North Sea, where not the 2 kb but the MiFish primer pair detected more species. This study demonstrates the feasibility of using long-read metabarcoding for eDNA vertebrate biodiversity assessments. However, our findings suggests that longer fragments are less abundant in environmental settings, but not in aquarium settings, suggesting that longer fragments may provide a more recent snapshot of the community. Thus, long-read metabarcoding can expand the molecular toolbox for biodiversity assessments by improving species-level identification and may be especially useful when the temporal origin of the eDNA signal is better understood.