Modern sequence analysis methods allow marker gene sequences to be resolved down to single nucleotide variants variously known as Oligotypes, Exact Sequence Variants (ESVs), or Amplicon Sequence Variants (ASVs). Although the naming conventions and algorithms may differ, they all share a common goal of deriving the true biological sequence from noisy data. In this protocol, a general pipeline for determining eASVs (exact amplicon sequence variants) is presented that is suitable for amplicon datasets containing mixed 16S\/18S sequences. We have chosen to use the open-source qiime2 framework for this analysis and have added one step at the beginning to split our mixed 16S\/18S amplicon libraries using the bbsplit.sh script from the versatile BBtools suite. In our experience, both deblur and DADA2 gave similar results when benchmarked against known sequences in mock communities. Both were able to resolve amplicon sequences from mock communities down to the exact (known) sequences, although DADA2 did produce spurious eASVs under certain circumstances. These spurious eASVs were, however, confined to situations where we knew that sequence data was generated in a non-standard way. Although deblur did not suffer from this same bias, its major disadvantage for our use case was that it discarded too much data (especially for our 18S sequences where up to 90% of sequences would be discarded vs only ~20% for DADA2). Therefore, we chose to use DADA2 with careful validation of the output as described in the protocol as our default pipeline.