Circulating microRNAs (ci-miRNAs) in blood have emerged as promising diagnostic, prognostic and predictive biomarkers in cancer. Many clinical studies currently incorporate studies that assess ci-miRNAs. Validation of the clinical significance of candidate biomarker miRNAs has proven to be difficult, potentially resulting from vast discrepancies in the detection methodology as well as biological variability. In the current study, the influence of several methodological factors on ci-miRNA detection was evaluated as well as short-term biological variability in patients with head and neck cancer. RNA was isolated from 124 serum and plasma samples originating from patients with head and neck cancer and healthy volunteers. The miRNA levels were measured using RT-qPCR and the influence of pre-analytical factors, different normalization strategies and temporal reproducibility was assessed. RNA carriers improved ci-miRNA detection in serum and plasma specimens. A prolonged pre-processing time correlated with an increased hemolytic index in serum samples only. Hemolysis differentially affected the detection of individual miRNAs. Optimal normalization was achieved using the averaged detection values of spike-in cel-miR-39-3p and endogenous miR-16-5p. Comparing biological replicates from patients with head and neck cancer, the intra-individual miRNA levels were relatively stable (average interval 7 days). Differences in the ci-miRNA detection methodology and limitations of currently used technologies can greatly affect the results and may explain inconsistent outcomes between studies. Prior to the implementation of ci-miRNAs as useful clinical biomarkers, further advances in the standardization of the detection methodology and reduction of technical variability are needed.