Background: Single-dose nevirapine (NVP) prophylaxis to mother and infant is widely used in resource-constrained settings for preventing mother-to-child transmission (MTCT) of HIV-1. Where women do not access antenatal care or HIV testing, post-exposure prophylaxis to the infant may be an important preventative strategy. Methods: This multicentre, randomized, open-label clinical trial (October 2000 to September 2002) in South Africa compared single-dose NVP with 6 weeks of zidovudine (ZDV), commenced within 24 h of delivery among 1051 infants whose mothers had no prior antiretroviral therapy. HIV-1 infection rates were ascertained at birth, and at 6 and 12 weeks of age. Kaplan-Meier survival methods were used to estimate HIV-1 infection rates in an intention-to-treat analysis. Results: Overall, 6 week and 12 week MTCT probability was 12.8% [95% confidence interval (Cl), 10.5-15.0] and 16.3% (95% Cl, 13.4-19.2), respectively. At 12 weeks, among infants who were not infected at birth, 24 (7.9%) infections occurred in the NVP arm and 41 (13.1%) in the ZDV arm (log rank P = 0.06). Using multivariate analysis, factors associated with infection following birth were ZDV use [odds ratio (OR), 1.8; 95% Cl, 1.1-3.2; P = 0.032), maternal CD4 cell count < 500 × 106 cells/l (OR, 2.5; 95% Cl, 1.3-5.0; P = 0.007), maternal viral load > 50 000 copies/ml (OR, 3.6; 95% Cl, 2.0-6.2; P < 0.0001) and breastfeeding (OR, 2.2; 95% Cl, 1.3-3.8; P = 0.006). Conclusion: A single-dose of NVP given to infants offers protection against HIV-1 infection and should be a strategy used in infants of mothers with untreated HIV infection. © 2005 Lippincott Williams & Wilkins.