AbstractFungi are crucial organisms in most ecosystems as they exert ecological key functions and are closely associated with land plants. Fungal community changes may, therefore, help reveal biodiversity changes in past ecosystems. Lake sediments contain the DNA of organisms in the catchment area, which allows reconstructing past biodiversity by using metabarcoding of ancient sedimentary DNA. We re‐evaluated various commonly used metabarcoding primers, and we developed a novel PCR primer combination for fungal metabarcoding to produce a short amplicon, thus accounting for amplification bias due to the degradation of ancient DNA. In silico PCRs showed higher diversity using this new primer combination, compared with previously established fungal metabarcoding primers. We analyzed data from sediment cores from four artic and one boreal lake in Siberia. These cores had been stored for 2–22 years after coring; we, therefore, examined the degradation effects of ancient DNA and storage time‐related bias affecting fungal communities. Amplicon lengths showed considerable variation within and between the major divisions of fungi, for example, amplicons of Basidiomycota were significantly longer than those of Mucoromycota; however, we observed no significant effect of sample age on amplicon length and GC content, suggesting the robustness of our results. We also found no indication of post‐coring fungal growth during storage regarding the proportions of common mold taxa, which would otherwise distort conclusions on past fungal communities. Terrestrial soil fungi, including mycorrhizal fungi and saprotrophs, were predominant in all lakes, whereas typical aquatic taxa were only represented to a negligible extent, which supports the use of lake sedimentary ancient DNA for reconstructing terrestrial communities.