In my opinion, the study has some limitations. First, CMV DNA detection in serum was performed by using the Cobas Amplicor CMV Monitor test (Roche Inc; Basel, Switzerland) (detection threshold, 400 CMV DNA copies/mL). I understand that the ultrasensitive version of the assay (limit of detection, 45 CMV DNA copies/mL) was not used. We previously showed that ICU patients with active CMV infection displayed low levels of plasma CMV DNAemia (median, 67 copies/mL, with CMV DNA loads > 400 copies/mL observed in ∼ 15% of samples), as determined by a real-time PCR assay (Abbott CMV PCR kit; Abbott Molecular; Des Plaines, Illinois), with a limit of detection of ∼ 10 copies/mL.3 Thus, the actual incidence of CMV DNAemia in the cohort of Gilbert et al4 might have been underestimated. Second, studies assessing the incidence and clinical relevance of active CMV infection in ICU patients should primarily include patients who are CMV-seropositive at the time of admission, as reactivation of latent infection rather than acquisition of primary infection is the usual mechanism underlying active CMV infection. Only 51% of patients had CMV-specific IgGs in the study by Gilbert et al.4 Third, the mean length of patients’ hospitalization in the ICU was 20.9 days.4 Earlier studies demonstrated that a relevant fraction of episodes of active CMV infection may develop after 3 weeks of ICU stay.1-3 Fourth, in a previous study,3 CMV reactivation was diagnosed in around 25% of ICU patients solely on the basis of the presence of CMV DNA in tracheal aspirates, indicating that monitoring for the presence of CMV in the blood compartment may underestimate the incidence of active CMV infection in this clinical setting.