BACKGROUND: In human body fluids, microRNA (miRNA) can be found as circulating cell-free miRNA (cfmiRNA), as well as secreted into extracellular vesicles (EVmiRNA). miRNAs are being intensively evaluated as minimally invasive liquid biopsy biomarkers in patients with cancer. The growing interest in developing clinical assays for circulating miRNA necessitates careful consideration of confounding effects of preanalytical and analytical parameters. METHODS: By using reverse transcription quantitative real-time PCR and next-generation sequencing (NGS), we compared extraction efficiencies of 5 different protocols for cfmiRNA and 2 protocols for EVmiRNA isolation in a multicentric manner. The efficiency of the different extraction methods was evaluated by measuring exogenously spiked cel-miR-39 and 6 targeted miRNAs in plasma from 20 healthy individuals. RESULTS: There were significant differences between the tested methods. Although column-based extraction methods were highly effective for the isolation of endogenous miRNA, phenol extraction combined with column-based miRNA purification and ultracentrifugation resulted in lower quality and quantity of isolated miRNA. Among all extraction methods, the ubiquitously expressed miR-16 was represented with high abundance when compared with other targeted miRNAs. In addition, the use of miR-16 as an endogenous control for normalization of quantification cycle values resulted in a decreased variability of column-based cfmiRNA extraction methods. Cluster analysis of normalized NGS counts clearly indicated a method-dependent bias. CONCLUSIONS: The choice of plasma miRNA extraction methods affects the selection of potential miRNA marker candidates and mechanistic interpretation of results, which should be done with caution, particularly across studies using different protocols.