This study aims to evaluate the diagnostic utilities of four leukocyte surface antigens-two lymphocyte antigens (CD25 and CD45RO) and two neutrophil antigens (CD11b and CD64)-for identification of late-onset nosocomial bacterial infection in preterm, very low birthweight infants, and to define the optimal cutoff value for each marker so that it may act as a reference with which future studies can be compared. Very low birthweight infants in whom infection was suspected when they were >72 h of age were eligible for the study. A full sepsis screen was performed in each episode. IL-6, C-reactive protein, and leukocyte surface antigens (CD25, CD45RO, CD11b, and CD64) were measured at 0 (at the time of sepsis evaluation), 24, and 48 h by standard biochemical methods and quantitative flow cytometric analysis. The diagnostic utilities including sensitivity, specificity, and positive and negative predictive values of each marker and combination of markers for predicting late-onset neonatal infection were determined. One hundred twenty-seven episodes of suspected clinical sepsis were investigated in 80 infants. Thirty-seven episodes were proven infection. The calculated optimal cutoff values for CD25, CD45RO, CD11b, and CD64 were 3,100, 2,900, 10,450, and 4,000 phycoerythrin-molecules bound per cell, respectively. An interim analysis of data after 68 episodes suggested that CD25 and CD45RO were poor predictors of neonatal infection with sensitivity or specificity <75% during a single measurement. Thus, these two markers were excluded from further investigation. In the final analysis, CD64 has the highest sensitivity (95-97%) and negative predictive value (97-99%) at 0 and 24 h after the onset. The addition of IL-6 or C-reactive protein (0 h) to CD64 (24 h) further enhanced the sensitivity and negative predictive value to 100%, and has the specificity and positive predictive value exceeding 88% and 80%, respectively. Neutrophil CD64 expression is a very sensitive marker for diagnosing late-onset nosocomial infection in very low birthweight infants. If further validated, the use of CD64 as an infection marker should allow early discontinuation of antibiotic treatment at 24 h without waiting for the definitive microbiologic culture results. The quantitative flow cytometric analysis applied in this study could be developed into a routine clinical test with high comparability and reproducibility across different laboratories.