The feasibility of the major peripheral blood leukocyte (PBL) subsets for use in qualitative and quantitative PCR to monitor secondary cytomegalovirus (CMV) infection and ganciclovir therapy was assessed with 188 blood samples derived from 40 CMV immunoglobulin G-positive renal-allograft recipients. In pp65 antigen-positive patients all leukocyte fractions, but only 79.5% of plasma preparations, were PCR positive. In pp65 antigen-negative samples from patients after antiviral treatment only 7.3% of polymorphonuclear cell (PMNL) samples, but 81.8% of peripheral blood mononuclear cells (PBMC), and 10.9% of plasma samples remained PCR positive. Similarly, in patients with latent infections only 5.0% of PMNL, but 51.7% of PBMC preparations, and 8.0% of plasma samples were PCR positive. Regarding patients with active CMV infection, CMV DNA copy numbers in PMNL correlated significantly with pp65 antigen-positive cell counts before and after onset of ganciclovir therapy. Significant differences in CMV DNA copy numbers in PMNL and plasma were observed (i) between patients with symptomatic infection and those with asymptomatic infection and (ii) between patients with active infection and those with latent infection. In contrast, PBMC harbored equally low CMV DNA levels both in patients with active infection and those with latent infections, and no decline of CMV DNA load in PBMC was observed during antiviral treatment. We conclude that detection of CMV DNA in PMNL, not in PBMC, is associated with active infections and is more sensitive than detection of CMV DNA in plasma. Negative PCR results for PMNL after antiviral therapy indicate recovery, and fewer unwanted positive results occur compared to PBMC and plasma. Therefore, purified PMNL should be preferred for analysis by qualitative CMV PCR to avoid unwanted positive results. The CMV DNA load in PBMC compared with that in PMNL is negligible during active infection, so mixed PBL are sufficient for use in quantitative PCR.