The objective of this paper is to evaluate the effect of different prevention strategies on the rate of early-onset neonatal group B streptococcus (GBS) disease and mortality. We compared the neonatal mortality and morbidity rates associated with early-onset GBS disease in three periods characterized by different prevention strategies, including no screening for GBS during pregnancy and no standardized chemoprophylaxis (1/1987 to 12/1990), antibiotic prophylaxis only with risk factors for GBS (1/1991 to 12/1994), and universal screening for GBS with rectovaginal cultures and chemoprophylaxis for women with positive results or risk factors (1/1995 to 12/1999). Statistical analysis included Fisher's exact test and Chi-square, with a two-tailed p <0.05 considered significant. The yearly prevalence of positive GBS cultures was similar throughout the screening period (mean 18%, range 16 to 19%). Compared with the no prophylaxis group (rate = 4/8,573), introduction of universal screening (rate = 0/13,754, p = 0.02) but not of prophylaxis for risk factors alone (rate = 1/10,303, p = 0.18) significantly decreased the occurrence of GBS-specific neonatal mortality. Universal screening decreased, though not significantly, the GBS-specific neonatal morbidity rates compared with a policy based on risk factors alone (0.4/1000 vs. 0.8/1000, p = 0.29). Our study had a power to detect a 0.7/1000 difference in the rate of specific morbidity between the two chemoprophylaxis policies (alpha = 0.05, beta= 0.80). Intrapartum prophylaxis for GBS, using universal screening or risk factors, is associated with a significant reduction in the specific neonatal mortality rate compared with no prophylaxis. Universal screening for GBS leads to a decrease in specific GBS morbidity compared with screening using risk factors alone.