Group B streptococci (GBS) are the leading cause of life-threatening neonatal bacterial infections in developed countries. The newborn is initially colonised during passage through the birth canal. Maternal vaginal carriage is usually asymptomatic. How safe and effective are strategies aimed at preventing severe neonatal GBS infection? To answer these questions, we conducted a review of the literature using the standard Prescrire methodology. In France, Group B streptococci are present in the vagina of about 10-15% of women towards the end of pregnancy. Mother-to-child transmission can occur when the membranes rupture, or during delivery. About one-third of infants born to mothers who carry Group B streptococci are colonised at birth. GBS colonisation rarely affects the newborn's health but, during the first 7 days of life, about 3% of colonised children develop a serious early-onset infection, particularly meningitis, which may be fatal or leave sequelae. Late-onset infections (after 7 days of life) do not appear to be linked to intrapartum colonisation. The risk of early neonatal GBS infection increases in cases of preterm delivery, maternal fever during delivery, and membrane rupture more than 18 hours before delivery. These situations account for 50% to 75% of early neonatal GBS infections. Several randomised trials suggest that intravenous antibiotic prophylaxis in women who carry Group B streptococci, from the onset of labour until delivery, reduces the risk of early-onset neonatal GBS infection from 4.7% to 0.4% (p = 0.02). Other antibiotic strategies are less well assessed or appear to be less effective. Penicillin G (benzylpenicillin) is the antibiotic of choice, while penicillin A (ampicillin or amoxicillin) is an alternative. In case of penicillin allergy, erythromycin or clindamycin are generally active against Group B streptococci and carry no particular risks for the infant. The greatest risk associated with penicillin, especially injectable forms, is an anaphylactic reaction, which can have severe consequences for both mother and child. The estimated frequency is about 5 cases per 10 000 treatments. The possible long-term adverse effects of antibiotic exposure during delivery are poorly documented in children. We found no randomised trials of standard vaginal screening for Group B streptococci between 35 and 38 weeks of gestation, or of a rapid PCR (polymerase chain reaction) test at the time of delivery. In a retrospective study conducted in the US, screening followed by antibiotic therapy if the results were positive was associated with a lower risk of early neonatal infection, but the methodology does not allow firm conclusions to be drawn. The risk-benefit balance has not been determined in terms of neonatal mortality. Widespread implementation of screening guidelines coincided with a decline in the incidence of early neonatal GBS infections in several countries. In the United Kingdom, systematic antibiotic therapy is recommended in high-risk situations, without systematic screening for Group B streptococci. In practice, the first priority is to identify situations in which there is a high risk of neonatal GBS infection, and to administer antibiotics during labour, after screening for GBS carriage, if possible. Outside of these situations, the risk of an anaphylactic reaction must be minimised by choosing the prophylactic antibiotic based on maternal allergy history, and by avoiding antibiotic prophylaxis altogether if the mother has a history of anaphylaxis, whatever the cause.