Phytoplankton play a vital role as primary producers in freshwater ecosystems, contributing to the nutrient cycle, energy flow, and ecological stability. To accurately assess phytoplankton diversity and community composition, this study compared traditional microscopy and environmental DNA (eDNA) metabarcoding in six small lakes located in the Han, Geum, and Nakdong River basins in Korea. eDNA analysis identified 268 species from 161 genera, approximately 2.4 times higher than microscopy, which detected 113 species from 68 genera. The eDNA data were dominated by picocyanobacteria such as Synechococcus and Cyanobium, while microscopy primarily revealed larger taxa, including Stephanodiscus and Scenedesmus. Nonmetric multidimensional scaling (NMDS) based on Bray-Curtis similarity showed clear separation between the two methods, with average similarity values of 0.0326 (1st survey) and 0.0221 (2nd survey) at the species level. Only 6.8% of the 429 total species were commonly detected by both methods, while overlap at the genus level was 18.8%. Spatial heterogeneity in phytoplankton communities based on eDNA was also evident depending on the sampling location, with the centre of the surface showing the highest species richness and overlap, suggesting its suitability for biodiversity monitoring. These findings demonstrate the high resolution and sensitivity of eDNA metabarcoding in capturing phytoplankton diversity and highlight its complementary role in existing biomonitoring programmes. Further improvements in the quantitative reliability of eDNA-based assessments will require efforts such as copy number normalisation, methodological standardisation, and refinement of reference databases.