BACKGROUND: Therapies containing two reverse transcriptase inhibitors (RTI) with or without protease inhibitors are used with increasing frequency in pregnant HIV-infected women. OBJECTIVE: To assess the safety of antiretroviral therapy in pregnant women and their newborns. METHODS: All clinical events and laboratory abnormalities in pregnant women on RTI with or without protease inhibitors and in their newborns were collected through an observational study. RESULTS: A total of 37 HIV-infected pregnant women have given birth to 30 children (by 30 April 1998). All received RTI, which were combined with protease inhibitors in 16 cases. Twelve women became pregnant while on treatment. Drugs used were as follows: zidovudine (n = 33), lamivudine (n = 33), stavudine (n = 4), indinavir (n = 9), ritonavir (n = 4), nelfinavir (n = 2) and saquinavir (n = 2). Adverse events during pregnancy were anaemia (n = 15), elevation of transaminases (n = 4), nausea/vomiting (n = 4), glucose intolerance (n = 2), nephrolithiasis (n = 2), diarrhoea (n = 2), hypertension (n = 1), insulin-requiring diabetes (n = 1). Adverse events in neonates were prematurity (n = 10), anaemia (n = 8), cutaneous angioma (n = 2), cryptorchidism (n = 2), transient hepatitis (n = 1). Non-life-threatening intracerebral haemorrhage occurred in a premature baby (33 weeks gestation) exposed during fetal life to zidovudine-lamivudine-indinavir, and in a term baby exposed to stavudine-lamivudine-indinavir. Extrahepatic biliary atresia occurred in one newborn exposed to zidovudine-lamivudine-indinavir. Maternal viral load was below 400 copies/ml in 18 out of 30 patients who delivered. One case of mother-to-child HIV transmission was identified. CONCLUSIONS: In HIV-infected pregnant women treated with two RTI with or without protease inhibitors, one or more adverse events occurred in 29 out of 37 women and in 14 out of 30 babies. In newborns, frequent prematurity, one case of biliary malformation and one intracerebral haemorrhage in a term baby are of concern. These observations do not preclude combination therapies during pregnancy but emphasize the necessity to maintain updated registers on their safety.