Epstein-Barr virus (EBV) DNA load monitoring in peripheral blood has been shown to be a useful tool for the diagnosis of aberrant EBV infections. In the present study we compared the relative diagnostic values of EBV DNA load monitoring in unfractionated whole blood and simultaneously obtained serum or plasma samples from Burkitt's lymphoma (BL) patients, transplant recipients, human immunodeficiency virus (HIV)-infected individuals, and infectious mononucleosis (IM) patients by a quantitative competitive PCR (Q-PCR). The EBV DNA load in BL patients was mainly situated in the cellular blood compartment (up to 4.5 x 10(6) copies/ml). EBV DNA loads in unfractionated whole blood and parallel serum samples showed no correlation. In transplant recipients, IM patients, and HIV-infected patients, the EBV burden in the circulation was almost exclusively restricted to the cellular blood compartment, because serum or plasma samples from these patients yielded negative results by Q-PCR, despite high viral loads in corresponding whole-blood samples. A 10-fold more sensitive but qualitative BamHI-W-repeat PCR occasionally revealed the presence of EBV at <2,000 copies of EBV DNA per ml of serum. Spiking of 100 copies of EBV DNA in samples with negative Q-PCR results excluded the presence of inhibitory factors in serum or plasma that influenced the Q-PCR result. Serum samples from all populations were often positive for beta-globin DNA, indicating cell damage in vivo or during serum preparation. We conclude that serum is an undesirable clinical specimen for EBV DNA load monitoring because it omits the presence of cell-associated virus and uncontrolled cell lysis may give irreproducible results or overestimation of the DNA load. Unfractionated whole blood is strongly preferred since it combines all blood compartments that may harbor EBV and it best reflects the absolute viral burden in the patient's circulation.