The application of high-throughput sequencing to retrieve multi-taxon DNA from different substrates such as water, soil, and stomach contents has enabled species identification without prior knowledge of taxon compositions. Here we used three minibarcodes designed to target mitochondrial COI in plankton, 16S in fish, and 16S in crustaceans, to compare ethanol- and tissue-derived DNA extraction methodologies for metabarcoding. The stomach contents of pygmy devilrays (Mobula kuhlii cf. eregoodootenkee) were used to test whether ethanol-derived DNA would provide a suitable substrate for metabarcoding. The DNA barcoding assays indicated that tissue-derived operational taxonomic units (OTUs) were greater compared to those from extractions performed directly on the ethanol preservative. Tissue-derived DNA extraction is therefore recommended for broader taxonomic coverage. Metabarcoding applications should consider including the following: (i) multiple barcodes, both taxon specific (e.g., 12S or 16S) and more universal (e.g., COI or 18S) to overcome bias and taxon misidentification and (ii) PCR inhibitor removal steps that will likely enhance amplification yields. However, where tissue is limited or no longer available, but the ethanol-preservative medium is still available, metabarcoding directly from ethanol does recover the majority of common OTUs, suggesting the ethanol-retrieval method could be applicable for dietary studies. Metabarcoding directly from preservative ethanol may also be useful where tissue samples are limited or highly valued; bulk samples are collected, such as for rapid species inventories; or mixed-voucher sampling is conducted (e.g., for plankton, insects, and crustaceans).