We sought to determine the efficacy of a 4-6 week course of zidovudine (ZDV) in a group of infants exposed to HIV-1. A retrospective chart review was conducted on HIV-1-exposed neonates identified from February 1998 to August 1999. These infants received ZDV and their mothers were counselled regarding the risks and benefits of breastfeeding. After informed consent was obtained, the HIV-1-status of the infant was determined by RNA-PCR after 1 month of age. Thirty-three HIV-1-exposed neonates were identified; seven infants were excluded, five because of insufficient data and two because of maternal ZDV administration. In all but three, the diagnosis was suspected because of maternal illness, 19 of 26 mothers having either stage 3 or 4 disease. The transmission rate was 15.4 per cent (4 of 26). Errors in ZDV administration were detected in 20 neonates of whom four were infected (p = 1; Fischer's exact test). Antenatal diagnosis of HIV-1 status was associated with fewer medication errors in HIV-1-exposed neonates (p = 0.017, Fisher's exact test). It was concluded that a transmission rate of 15.4 per cent in symptomatic mothers suggests efficacy. Under similar circumstances, transmission rates varying between 37 and 67 per cent have been reported. Antenatal diagnosis is significantly associated with compliance to the regimen. Early identification of at-risk neonates and familiarity with the regimen may improve the outcome.