Coastal marine fish populations are in decline due to overfishing, habitat destruction, climate change and invasive species. Seasonal monitoring is important for detecting temporal changes in the composition of fish communities, but current monitoring is often non-existent or limited to annual or semi-annual surveys. In the present study, we investigate the potential of using environmental DNA (eDNA) metabarcoding of seawater samples to detect the seasonal changes in a coastal marine fish community. Water sampling and snorkelling visual census were performed over 1 year (from 23rd of August 2013 to 11th of August 2014) at a temperate coastal habitat in Denmark (55°45′39″N, 12°35′59″E) and compared to long-term data collected over a 7-year period. We used Illumina sequencing of PCR products to demonstrate that seawater eDNA showed compositional changes in accordance with seasonal changes in the fish community. The vast majority of fish diversity observed in the study area by snorkelling was recovered from sequencing, although the overlap between methods varied widely among sampling events. In total, 24 taxa were detected by both methods, while five taxa were only detected using eDNA and three taxa were only detected by snorkelling. A limitation of the applied primers was the lack of resolution to species level in a few diverse families, and varying sequencing depth between samples represents a potential bias. However, our study demonstrates the utility of eDNA for recovering seasonal variation in marine fish communities, knowledge of which is essential for standardised long-term monitoring of marine biodiversity.