1. DNA metabarcoding has developed into a commonly used tool for biodiversity assessment and monitoring. How results from DNA metabarcoding are compared with studies based on ‘classic’, in most cases morphological species identification, is still unclear. Studies investigating species detection against a known baseline are virtually non-existent. 2. In this study, we used light trap samples collected in eastern Austria to investigate the concordance between morphological species lists and results obtained from COI metabarcoding using the Illumina MiSeq platform. Two primer combinations of different lengths (313 and 205 bp) were compared to assess the influence of amplicon length. 3. Species detection rates ranged between 0.38 and 0.69; the shorter amplicon had on average higher species detection rates compared with the longer amplicon. Singleton species were less likely to be detected through metabarcoding. 4. The major determinant for a species to be detected was its biomass, viz. smaller species had a lower chance to be detected. However, there is also evidence of taxonomic bias on the level of superfamilies. While the influence of biomass is to be expected, the presence of taxonomic bias gives reason for concern and requires further studies. Such a bias can be of significance when metabarcoding is used to determine conservation measures. Ordination analyses of all sampling sites showed that as far as community ecology is concerned, the overall pattern obtained from the full species list was mostly preserved in our metabarcoding results.