DNA metabarcoding combined to high-throughput sequencing (HTS) is a promising approach for freshwater biomonitoring using benthic diatoms. This molecular approach is becoming faster, cheaper and allows handling more samples than the classical morphological one. rbcL gene has been proven suitable as a barcode for diatoms providing qualitatively relevant taxonomic inventories; however, optimizations are still required for quantifying relative species abundances. We hypothesized that rbcL copy number is correlated to diatom cell biovolume (as it is the case for 18S gene) and a correction factor that account for cell biovolume is applicable to optimize taxa quantification. An experiment was carried out using 8 pure diatom cultures chosen for their contrasted biovolumes. The 18S and the rbcL copy numbers were both correlated to diatom cell biovolume, but only rbcL gene followed a linear model applicable to the 8 species, thus allowing to define a correction factor. The efficiency of the correction factor was tested on 5 mock communities composed by different proportions of DNA from the 8 species. The 5 mock communities were sequenced using HTS by targeting the rbcL barcode, and the obtained proportions of DNA reads were corrected using the defined model. The correction made molecular inventories closer to the morphological ones. This correction factor was finally validated on environmental DNA metabarcoding data (biofilms) and was shown to reduce differences between molecular and morphological water quality indices. Our results show that the correction factor is effective for correcting quantitative biases in diatoms metabarcoding studies, improving the resulting water quality assessment.