Summary The application of metabarcoding to study animal‐associated microeukaryotes has been restricted because the universal barcode used to study microeukaryotic ecology and distribution in the environment, the Small Subunit of the Ribosomal RNA gene (18S rRNA), is also present in the host. As a result, when host‐associated microbial eukaryotes are analysed by metabarcoding, the reads tend to be dominated by host sequences. We have done an in silico validation against the SILVA 18S rRNA database of a non‐metazoan primer set (primers that are biased against the metazoan 18S rRNA) that recovers only 2.6% of all the metazoan sequences, while recovering most of the other eukaryotes (80.4%). Among metazoans, the non‐metazoan primers are predicted to amplify 74% of Porifera sequences, 4% of Ctenophora, and 15% of Cnidaria, while amplifying almost no sequences within Bilateria. In vivo , these non‐metazoan primers reduce significantly the animal signal from coral and human samples, and when compared against universal primers provide at worst a 2‐fold decrease in the number of metazoan reads and at best a 2800‐fold decrease. This easy, inexpensive, and near‐universal method for the study of animal‐associated microeukaryotes diversity will contribute to a better understanding of the microbiome.