Basic knowledge on the biology and epidemiology of equine strongylid species still needs to be improved to contribute to the design of better parasite control strategies. Nemabiome metabarcoding is a convenient tool to quantify and identify species in bulk samples that could overcome the hurdle that cyathostomin morphological identification represents. To date, this approach has relied on the internal transcribed spacer 2 (ITS-2) of the ribosomal RNA gene, with a limited investigation of its predictive performance for cyathostomin communities. Using DNA pools of single cyathostomin worms, this study aimed to provide the first elements to compare performances of the ITS-2 and a cytochrome c oxidase subunit I (COI) barcode newly developed in this study. Barcode predictive abilities were compared across various mock community compositions of two, five and 11 individuals from distinct species. The amplification bias of each barcode was estimated. Results were also compared between various types of biological samples, i.e., eggs, infective larvae or adults. Bioinformatic parameters were chosen to yield the closest representation of the cyathostomin community for each barcode, underscoring the need for communities of known composition for metabarcoding purposes. Overall, the proposed COI barcode was suboptimal relative to the ITS-2 rDNA region, because of PCR amplification biases, reduced sensitivity and higher divergence from the expected community composition. Metabarcoding yielded consistent community composition across the three sample types. However, imperfect correlations were found between relative abundances from infective larvae and other life-stages for Cylicostephanus species using the ITS-2 barcode. While the results remain limited by the considered biological material, they suggest that additional improvements are needed for both the ITS-2 and COI barcodes.