OBJECTIVE: In a health maintenance organization population, we determined the incidence of early-onset (at less than or equal to 7 days) neonatal group B streptococcal (GBS) disease, the sensitivity and prevalence of labor risk factors, the adherence to a protocol for intrapartum antibiotics, and the costs for care of and outcomes of affected infants. METHODS: Mothers and infants at four health maintenance organization hospitals in northern California in 1989 to 1995 were studied retrospectively using computerized databases and chart review. In 1994, two of the four hospitals had adopted protocols similar to the ACOG recommendations for intrapartum antibiotics for women with labor risk factors (preterm, temperature 100.4F or higher, or rupture of membranes (ROM) 18 hours or more). RESULTS: Among the 79,940 live births, the incidence of early-onset neonatal GBS infection was higher among preterm than among term infants (3.1 compared with 0.9 per 1000). Before protocol adoption, 68% of 65 infants with GBS had mothers with labor risk factors. Approximately 18% of all mothers had labor risk factors: 7.7% had preterm delivery, and 10.6% had term delivery with fever and/or ROM 18 hours or more. At the two hospitals that adopted GBS protocols, GBS incidence was reduced from 1.3 per 1000 in the preprotocol period to 0.8 per 1000 in the postprotocol period (P=.08). Six cases of neonatal GBS occurred after protocol adoption. Of these, four were not preventable under the protocol and two might have been preventable if protocol had been followed. Three of the 19 preterm infants with group B streptococcal disease died. CONCLUSION: Risk factor-based protocols hold some promise to reduce GBS disease, but clinical strategies to promote protocol adherence are needed.