Abstract Marine sponges have recently been recognised as natural samplers of environmental DNA (eDNA) due to their effective water filtration and their ubiquitous, sessile and regenerative nature. However, laboratory workflows for metabarcoding of sponge tissue have not been optimised to ensure that these natural samplers achieve their full potential for community survey. We used a phased approach to investigate the influence of DNA isolation procedures on the biodiversity information recovered from sponges. In Phase 1, we compared three treatments of residual ethanol preservative in sponge tissue alongside five DNA extraction protocols. The results of Phase 1 informed which ethanol treatment and DNA extraction protocol should be used in Phase 2, where we assessed the effect of starting tissue mass on extraction success and whether homogenisation of sponge tissue is required. Phase 1 results indicated that ethanol preservative may contain unique and/or additional biodiversity information to that present in sponge tissue, but blotting tissue dry generally recovered more taxa and generated more sequence reads from the wild sponge species. Tissue extraction protocols performed best in terms of DNA concentration, taxon richness and proportional read counts, but the non-commercial tissue protocol was selected for Phase 2 due to cost-efficiency and greater recovery of target taxa. In Phase 2 overall, we found that homogenisation may not be required for sponge tissue and more starting material does not necessarily improve taxon detection. These results combined provide an optimised DNA isolation procedure for sponges to enhance marine biodiversity assessment using natural sampler DNA metabarcoding.