Cytomegalovirus (CMV) is an important cause of transfusion-associated morbidity and mortality; however, only 0.4 to 12% of the blood products obtained from seropositive blood donors transmit infection. The effects of three commercially available whole-blood sample preparation kits on the detection of CMV PCR products by a semiquantitative adaptation of the Digene SHARP Signal System Assay (DSSSA) in samples from volunteer blood donors was assessed. Of 101 samples from seropositive blood donors, CMV was detected in 0 (0%) of the samples extracted with a QIAamp blood kit (QIAGEN), 1 (1%) of the samples extracted with an Amplicor whole-blood specimen preparation kit (Roche), and 8 (8%) of the samples extracted with an Isoquick nucleic acid extraction kit (modified by the addition of carrier tRNA) (Microprobe). CMV DNA was not detected in samples from seronegative blood donors (n = 13). Nested PCR of selected samples confirmed the detection of CMV in the sane eight samples extracted with the modified Isoquick nucleic acid extraction kit and detected an additional nine CMV-positive samples (n = 50). Samples from volunteer blood donors contain low copy numbers of CMV DNA. PCR amplification of such specimens can result in analytical sampling errors, giving results similar to the variations in titers recognized during determinations of the 50% tissue culture infective dose. The detection of CMV in blood samples from volunteer blood donors by PCR is a function of sample preparation, amplification conditions, and detection methodology. Accurate assessments of the clinical utility of CMV DNA detection by nucleic acid amplification for blood product screening and patients will require highly standardized and quantitative methodology.