To determine the effect of single-dose penicillin given at birth on the rate of early-onset group B streptococcal (GBS) invasive disease in an inner-city population. Laboratory-based surveillance of GBS disease from 1972–1994 at Parkland Memorial Hospital and Children's Medical Center in Dallas, Texas, was reviewed retrospectively. All infants born at Parkland Memorial Hospital from January 1, 1972 to December 31, 1994, or a total of 259,049 live births, were included. Early-onset (within 3 days) GBS disease rates were compared for each of five observation groups to determine the efficacy of a single dose of aqueous penicillin G (50,000 U for infants weighing 2000 g or more and 25,000 U for those weighing less than 2000 g) administered intramuscularly within 1 hour of delivery for prevention of GBS disease. The rates of early-onset GBS disease were compared in five observation groups: A) pre-study, January 1, 1972 to December 3, 1977—no GBS prophylaxis; B) prospective, controlled intervention study, December 4, 1977 to May 31, 1981, including infants who received a single dose of penicillin at birth (group B1) and those who did not (group B2); C) universal penicillin prophylaxis, June 1, 1981 to October 31, 1986; and D) no routine penicillin prophylaxis, November 1, 1986 to December 31, 1994. The incidence of early-onset GBS disease in the penicillin groups (B1, C) was significantly lower than that in the untreated groups (A, B2, D): 0.25 and 0.63 per 1000 versus 1.59, 1.19, and 1.95 per 1000, respectively (P ≤ .03). The incidence of late-onset GBS disease was unaffected by penicillin prophylaxis, and there was no increase in the incidence of disease caused by penicillin-resistant pathogens or associated mortality in penicillin-treated infants: 2.2 and 2.1 per 1000 versus 1.6 and 3.3 per 1000 for disease; 1.0 and 0.5 per 1000 versus 0.4 and 0.3 per 1000 for deaths. Universal administration of single-dose penicillin at birth is a safe and effective intervention for the prevention of early-onset GBS disease.