We compared several previously defined scoring systems using white blood cell indices as part of a retrospective evaluation of infants with early onset Group B streptococcal (GBS) sepsis. Nineteen newborns were diagnosed with GBS sepsis between January, 1988, and April, 1990. Case controls (n = 33) were selected from patients admitted to the Neonatal Intensive Care Unit for suspected sepsis. Complete blood counts obtained at admission and between 12 and 24 hours of age were reviewed. There was a significant change in the ratio of immature to total neutrophils in the GBS group over time. Scoring systems for neonatal sepsis by Manroe et al., Rodwell et al. and Spector et al. had poor sensitivity, specificity, positive predictive value and negative predictive value when initial white blood cell count criteria were used, but scoring systems by Manroe and Rodwell were 100% sensitive and had 100% negative predictive value when applied to the repeat white blood cell count. We conclude that a single early complete blood count may not be an adequate screening tool for early onset GBS sepsis and should not be used to rule out infection. Optimal screening for GBS sepsis requires a repeat complete blood count within the first 24 hours of age.