BACKGROUND: With the widespread implementation of intrapartum antibiotic prophylaxis (IAP), the rate of early-onset neonatal sepsis and meningitis caused by Streptococcus agalactiae (group B streptococcus [GBS]) has decreased dramatically, especially in term infants. However, cases of GBS disease continue to occur despite IAP and incur significant morbidity and mortality. Inaccurate screening results, improper implementation of IAP, or antibiotic failure all may contribute to persistent disease. OBJECTIVE: To determine if clinical, procedural, or microbiologic factors influenced persistent early-onset GBS disease (EOGBS) cases in a single large maternity service after the institution of a screening-based protocol for IAP. METHODS: Retrospective review of all cases of culture-proven EOGBS at the Brigham and Women's Hospital (Boston, MA) from 1997 to 2003. Serotyping and surface protein analyses were performed on available disease isolates. RESULTS: A total of 67260 infants were live-born during this period. Twenty-five cases of EOGBS (0.37 of 1000 live births) were identified. The overall incidence of EOGBS progressively decreased with different approaches to IAP. Of the 25 cases identified after institution of a screening-based protocol, 17 (68%) occurred in term infants (1 death), and 8 (32%) occurred in preterm infants (3 deaths). Among the mothers of term infants, 14 of 17 (82%) had been screened GBS negative; 1 was GBS unknown. More than half of the mothers of term infants who had screened GBS negative (8 of 14) had intrapartum risk factors for neonatal infection but did not receive antibiotics before delivery. Ten of the 17 term infants were evaluated for infection because of clinical signs of illness, and the remainder were evaluated because of intrapartum sepsis risk factors. Of the mothers of preterm infants, by the time of delivery 3 of 8 had been documented as GBS positive, 2 of 8 had been documented GBS negative, and 3 of 8 remained unknown. Only 1 of 25 women received adequate IAP, but the isolate was resistant to the administered antibiotic (clindamycin). Antibiotic resistance was not a factor in any other case, and no dominant serovariant was identified among tested isolates. Procedural errors (lack of recognition of documented GBS colonization or failure to evaluate infants at risk for sepsis) were identified in 4 cases. CONCLUSIONS: The majority of the remaining cases of EOGBS occurred in infants whose mothers screened negative for GBS colonization. Even in the setting of a maternal GBS-screening program, efforts to evaluate and treat infants with intrapartum clinical risk factors for early-onset sepsis remain important. Until effective vaccines against GBS are available for clinical use, development and implementation of rapid and sensitive techniques for screening for GBS status and antibiotic susceptibility at presentation may help prevent additional cases of invasive GBS disease.