<p>To unravel complex plant-pollinator interactions, it is essential to be able to obtain not only qualitative but also quantitative information about the composition of pollen carried by insects. This information can be gathered by light microscopy identification of pollen grains, but this process can be time-consuming and requires trained specialists. ITS2 metabarcoding can be a powerful tool, providing a high taxonomic resolution in identifying taxa represented in a pollen sample; however, a possible critical point of this approach may be its quantitative reliability in estimating the relative abundance of pollen taxa, especially when ITS2 is used as a single marker. The aim of this work was to test the quantitative reliability of the ITS2 metabarcoding approach in assessing the composition of pollen carried by insects. Pollen samples consisting of a limited number of grains were retrieved from different bee families collected in different environments. The NGS reads obtained by ITS2 metabarcoding were compared at the family level with the pollen grain counts obtained by light microscopy. Moreover, the proportion of NGS reads corresponding to blooming plants occurring in the sampling area was evaluated at the genus level. The results highlighted a good correlation between the relative abundances obtained with metabarcoding and microscopy, and a good accuracy of metabarcoding in providing quantitative results independently of bee taxon and pollen taxa composition. Furthermore, this study indicates that ITS2 metabarcoding can be applied when the available amount of pollen is a constraint, an important point in the perspective of an actual evaluation of plant-pollinator interactions.</p>