OBJECTIVE: To evaluate the efficacy of antiretroviral therapies in reducing the risk of mother-to-child transmission of HIV infection. METHODS: Systematic review and meta-analysis of randomized controlled trials. Clinical trials of antiretrovirals were identified through electronic searches (MEDLINE, EMBASE, BIOSIS, EBM review and the Cochrane Library) up until November 2006. Historical searches of reference lists of relevant randomized controlled trials, and systematic and narrative reviews were also undertaken. Studies were included if they were (i) randomized controlled trials of any antiretroviral therapy aimed at decreasing the risk of mother-to-child transmission of HIV infection, (ii) reporting outcomes in terms of HIV infection in infant, infant death, stillbirth, premature delivery, or low birth weight. The data were extracted by a single investigator and checked by a second investigator. Disagreements were resolved through discussion or a third investigator. The efficacy was estimated using relative risk (RR), risk difference (RD) and number needed to treat (NNT) together with 95% confidence intervals. RESULTS: Fifteen trials were included in the systematic review. Based on five placebo-controlled trials, a zidovudine regimen reduced the risk of mother-to-child transmission by 43% (95% CI: 29-55%). The incidence of low birth weight seems to be decreased with zidovudine (pooled RR 0.75, 95% CI: 0.57-0.99). The efficacy of short-short course of zidovudine was comparable with that of the long-short course. Nevirapine monotherapy given to mothers and babies as a single dose reduced the risk of vertical transmission compared with an intrapartum and post-partum regimen of zidovudine (RR 0.60, 95% CI: 0.41-0.87). Zidovudine plus lamivudine was effective in reducing the risk of maternal-child transmission of HIV (RR 0.63, 95% CI: 0.45-0.90). Adding zidovudine to single-dose nevirapine in babies was no more effective than nevirapine alone (pooled RR 0.88, 95% CI: 0.47-1.63), nor was there any significant difference between zidovudine plus lamivudine and nevirapine. In mothers who were treated with standard antiretroviral therapy, no additional benefit was observed with the addition of a single dose of nevirapine in mothers and newborns. In addition, for mothers who received zidovudine prophylaxis, a two-dose intrapartum/newborn nevirapine reduced the risk of HIV infection and death of babies by 68% (95% CI: 39-83%) and 80% (95% CI: 10-95%), respectively, when compared with placebo. CONCLUSIONS: The available evidence suggests that zidovudine alone or in combination with lamivudine and nevirapine monotherapy is effective for the prevention of mother-to-child transmission of HIV. They may also be beneficial in reducing the risk of infant death. Different antiretroviral regimens appear to be comparably effective in reducing HIV transmission from mothers to babies. In mothers already receiving zidovudine prophylaxis, adding a single dose of nevirapine to mothers during labour and giving the same drug to infants may further decrease the risk of vertical transmission and infant death.