Environmental DNA (eDNA) is used to detect biodiversity by the capture, extraction, and identification of DNA shed to the environment. However, eDNA capture and extraction protocols vary widely across studies. This use of different protocols potentially biases detection results and could significantly hinder a reliable use of eDNA to detect biodiversity. We tested whether choice of eDNA capture and extraction protocols significantly influenced biodiversity detection in aquatic systems. We sampled lake and river water, captured and extracted eDNA using six combinations of different protocols with replication, and tested for the detection of four macroinvertebrate species. Additionally, using the same lake water technical replicates, we compared the effect of capture and extraction protocols on metabarcode detections of biodiversity using 16S for eubacteria and cytochrome c oxidase I (COI) for eukaryotes. Protocol combinations for capture and extraction of eDNA significantly influenced DNA yield and number of sequences obtained from next generation sequencing. We found significantly different detection rates of species ranging from zero percent to thirty-three percent. Differences in which protocol combinations produced the highest metabarcoded biodiversity were detected and demonstrate that different protocols are required for different biodiversity targets. Our results highlight that the choice of molecular protocols used for capture and extraction of eDNA from water can strongly affect biodiversity detection. Consideration of biases caused by choice of protocols should lead to a more consistent and reliable molecular workflow for repeatable and increased detection of biodiversity in aquatic communities.