OBJECTIVE To evaluate three strategies for preventing group B streptococcal neonatal sepsis in large teaching hospitals and to examine their cost effectiveness and cost benefit. METHODS A decision tree was constructed to determine the outcome of three prevention strategies: (1) Screen all pregnant women for group B streptococcus (GBS) at 28 weeks' gestation and give intrapartum chemoprophylaxis to all GBS carriers; (2) Screen all pregnant women, but give chemoprophylaxis selectively to those carriers with obstetric risk factors (i.e., premature labour, prolonged rupture of membranes and maternal sepsis); and (3) Do not screen but give intrapartum chemoprophylaxis for all women with obstetric risk factors. Australian data were used if available, and the cost benefit and cost effectiveness of each strategy compared with no screening at all were estimated. RESULTS All three strategies had a net cost benefit compared with no intervention at all. Strategy 1 could prevent 46%, Strategy 2 38% and Strategy 3 60%-80% of all early-onset GBS sepsis. CONCLUSIONS A substantial proportion of early-onset GBS sepsis is preventable. Strategy 1, which is practical in some large teaching hospitals, prevents more sepsis at a lower cost per case prevented than Strategy 2. Strategy 3 theoretically could prevent more cases at a substantially lower cost than Strategies 1 or 2, but has not been evaluated clinically.