Diatoms (Bacillariophyceae) are widely used as bioindicators of present and past water quality because they inhabit the vast majority of aquatic ecosystems, are very diverse, highly sensitive to a variety of environmental conditions, and are characterized by silicified cell walls that favor their long-term preservation in sediments. Alongside with traditional morphological analyses, metabarcoding has become a valuable tool to study the community structures of various organisms, including diatoms. Here, we aimed to test whether the quantity of sediment sample used for DNA extraction is affecting the results obtained from high-throughput sequencing (metabarcoding) of the diatom rbcL region by isolating DNA from 10 g and 0.5 g (wet weight) of lake surface sediment samples. Because bioinformatics processing of metabarcoding data may affect the outcome, we also tested the consistency of the results from three different pipelines. Additionally, the agreement between metabarcoding data and morphological inventories of corresponding samples were compared. Our results demonstrate highly uniform patterns between the diatom rbcL amplicons from 10 g and 0.5 g of DNA extracts (HTS 10 and HTS 0.5, respectively). Furthermore, metabarcoding results were highly consistent among the data sets produced by different bioinformatics pipelines. Comparing results from metabarcoding and microscopy, we identified some taxonomic mismatches, which are related to the common issue of incompleteness of the sequence databases, but also to inconsistencies in diatom taxonomy in general and potential dissolution effects of diatom valves caused by high alkalinity of the investigated lake waters. Nevertheless, multivariate community analysis demonstrated highly similar results between data sets identified by microscopy and metabarcoding, further confirming that metabarcoding is a viable alternative for identifying diatom-environment relationships.