Circulating microRNAs that are associated with specific diseases have garnered much attention for use in diagnostic assays. However, detection of disease-associated miRNA can be affected by several factors such as release of contaminating cellular miRNA during sample collection, variations due to amplification of transcript for detection, or controls used for normalization for accurate quantitation. We analyzed circulating miRNA in serum and plasma samples obtained concurrently from 28 patients, using a Nanostring quantitative assay platform. Total RNA concentration ranged from 32-125 μg/ml from serum and 30-220 μg/ml from plasma. Of 798 miRNAs, 371 miRNAs were not detected in either serum or plasma samples. 427 were detected in either serum or plasma but not both, whereas 151 miRNA were detected in both serum and plasma samples. The diversity of miRNA detected was greater in plasma than in serum samples. In serum samples, the number of detected miRNA ranged from 3 to 82 with a median of 17, whereas in plasma samples, the number of miRNA detected ranged from 25 to 221 with a median of 91. Several miRNA such as miR451a, miR 16-5p, miR-223-3p, and mir25-3p were highly abundant and differentially expressed between serum and plasma. The detection of endogenous and exogenous control miRNAs varied in serum and plasma, with higher levels observed in plasma. Gene expression stability identified candidate invariant microRNA that were highly stable across all samples, and could be used for normalization. In conclusion, there are significant differences in both the number of miRNA detected and the amount of miRNA detected between serum and plasma. Normalization using miRNA with constant expression is essential to minimize the impact of technical variations. Given the challenges involved, ideal candidates for blood based biomarkers would be those that are indifferent to type of body fluid, are detectable and can be reliably quantitated.