Circulating DNA isolated from serum and plasma has been shown to be a useful biomarker in various diseases including cancer. Serum reportedly contains a higher amount of free circulating DNA than it does in plasma. The underlying reason for this is unclear, but important because it may have clinical implications in interpreting results and using the appropriate resource. Twenty-four pairs of serum and plasma samples were collected from patients with tumors, and free circulating DNA was quantified by real-time quantitative PCR (qPCR) for the ALU repeats, which had a sensitivity of 0.1 pg/microL of DNA in serum/plasma. The possibility of DNA loss was eliminated because ALU-qPCR does not require DNA purification from serum/plasma. The DNA concentrations of serum and plasma samples were 970 +/- 730 pg/microL and 180 +/- 150 pg/microL (mean +/- SD), respectively. The amount of DNA in paired serum and plasma specimens was positively correlated (R = 0.72 and P = 0.0002). An estimated 8.2% of total DNA in serum was extraneous; the concentration of DNA was 6.1 +/- 3.5 (mean +/- SD)-fold higher in serum than in paired plasma after subtraction of it. Contribution of extraneous DNA from cells in blood ruptured during the separation step was minor for explaining the difference between serum and plasma. A possible explanation was unequal distribution of DNA during separation from whole blood. We advocate that serum is a better specimen source for circulating cancer-related DNA as a biomarker.