Positive results by cytomegalovirus (CMV) PCR of plasma are considered predictive of active CMV infection in kidney allograft recipients. To assess whether contamination with leukocyte-derived CMV DNA can distort the results, aliquots of whole-blood samples from 60 CMV immunoglobulin G-positive patients with leukocyte CMV DNAemia were stored for up to 24 h at room temperature (RT) and at 4°C before plasma preparation. Native and ultrafiltered plasma samples were tested by CMV and -globin PCRs. Among 30 latently infected patients (negative for CMV pp65 antigens), low baseline rates (10%) and levels (median number of copies, 10 [per 10 l]) of CMV plasma DNAemia in native plasma samples increased significantly over time (after 4 h at RT, 37% [P < 0.001]; median number of copies, 45 [P < 0.001]). Similar effects were found during storage at 4°C. Ultrafiltration reduced the levels of CMV plasma DNAemia, but by 6 h of storage the levels were significantly elevated as well. CMV and -globin DNA kinetics in plasma were parallel. In contrast, 30 actively infected patients (pp65 positive) had high baseline rates (87% in native samples) and levels (median number of copies, 75) of CMV plasma DNAemia. No significant effects of storage or ultrafiltration and no concordance with -globin DNA kinetics were seen. In conclusion, delayed preparation of plasma samples bears a significant risk of false-positive CMV PCR results, probably due to leukocyte lysis. This has important implications in the clinical setting and for PCR standardization.