OBJECTIVES: To assess the effectiveness and safety of zidovudine (ZDV) for preventing mother-to-child transmission (MTCT) of HIV. METHODS: A systematic review of randomized controlled trials (RCTs) was conducted using the methodology of The Cochrane Collaboration. PUBMED, EMBASE, CINAHL, AIDSearch, The Cochrane Library (Issue 4, 2007), AIDSinfo, CRD (Center of Review and Dissemination) databases and three Chinese Databases (CBM, CNKI, VIP) were searched from establishment to December 31, 2007. We also searched the documents of government and non-governmental organizations, as well as the abstracts from relevant conferences. RCTs assessing the effects of ZDV for preventing MTCT were included. Trial selection, quality assessment and data extraction were done by two reviewers independently, with confirmation by cross-checking the information. RESULTS:  Eight original studies, involving 24 published articles and 13 conference abstracts were included. All included studies were of high quality with Jadad score of 3 or more. Our meta-analyses showed that: Based on four trials (2385 participants), any ZDV regimen (both long course and short course, either in breast-feeding population or not) significantly reduced the risk of mother-to-child transmission of HIV compared with placebo by 43%-50%. The incidences of stillbirth, infant and maternal mortality, premature delivery, low birth weight, birth defects, and any adverse reactions (either in mothers or in children) were comparable between ZDV and placebo (P>0.05). One trial (1437 participants) compared different courses of ZDV. The “long-long” course (from 28 weeks in pregnancy for the mother and for the baby until 6 weeks old) decreased the risk of transmission by 61% (RR=0.39, 95%CI 0.19 to 0.82), compared to the “short-short” course of ZDV (from 35 weeks in pregnancy for the mother and for the baby until 3 days old). However, the effectiveness of the “long-short” course (from 28 weeks in pregnancy for the mother and for the baby until 3 days old) and the “short-long” course (from 35 weeks in pregnancy for the mother and for the baby until 6 weeks old) did not differ from that of the “long-long” course. There was no difference among the different courses of ZDV in the incidence of infant and maternal mortality, or any adverse events in mothers or children (P>0.05).  One trial (1 200 participants) compared formula-fed plus short course ZDV with breastfed plus long course ZDV. The formula-fed plus ZDV reduced the risk of transmission by 35%-39% in the period 7 months to 18 months after birth, and was associated with a higher mortality rate at 7 months than the breastfed plus ZDV group (9.3% vs. 4.9%; P=0.003). The incidences of infant and maternal mortality or any adverse reaction (either in mothers or in children) were similar between the two groups (P>0.05). Two RCTs (702 participants) showed that single dose nevirapine (sdNVP) given to mothers and babies was more effective than short course or ultra-short course regimens of ZDV (risk decreased by 44%-65%). The incidences of infant and maternal mortality, any adverse reactions in mothers or children were similar between the two groups (P>0.05). CONCLUSIONS: Compared with placebo, any ZDV course was more effective in preventing MTCT of HIV with a similar safety profile. The “long-long” course ZDV decreased the risk of transmission, when compared with “short-short” course of ZDV, but the effectiveness and safety of “long-long” course, “long-short” course or “short-long” course of ZDV were similar. The formula-fed plus short course ZDV reduced the risk of transmission, and was associated with a higher mortality rate at 7 months compared with the breastfed plus long course ZDV after birth. Single dose nevirapine was more effective than short course or ultra-short course ZDV and had a similar safety profile.