Objectives: To evaluate the effect of maternal, obstetric, neonatal and post-natal factors on the risk of vertical transmission of HIV-1. Design: Multicentre retrospective cohort study. Setting: Obstetric and paediatric clinics in four cities in Sao Paulo State, Brazil. Main outcome: Child's HIV-1 infection status. Methods: Data were collected by standardized record abstraction and interview on 553 children born to women identified as HIV-1-infected before or at delivery. Paediatric infection was determined by immunoglobulin G anti-HIV-1 tests at age 18 months or by AIDS diagnosis at any age. Multivariate logistic regression was used to assess the effect of potential risk factors on vertical transmission of HIV-1. Results: HIV-1 infection status was determined for 434 children (follow-up rate of 78%); 69 were classified as HIV-1-infected [transmission risk, 16%; 95% confidence interval (CI), 13-20%]. In multivariate analysis, advanced maternal HIV-1 disease [odds ratio (OR), 4.5; 95% CI, 2.1-9.5], ever breastfed (OR, 2.2; 95% CI, 1.2-4.2), child's negative Rhesus blood group (OR, 2.5; 95% CI, 1.2-5.5), third trimester amniocentesis (OR, 4.1; 95% CI, 1.2-13.5) and black racial group (OR, 0.3; 95% CI, 0.1-0.9) were independently and significantly associated with mother-to-child transmission of HIV-1. Transmission was increased marginally with prematurity, more than 10 lifetime sexual partners and prolonged duration of membrane rupture. No association was found between child's HIV-1 infection and mode of delivery or serological evidence of syphilis during pregnancy. Conclusion: These findings support the importance of severity of maternal HIV-1 disease in the risk of vertical transmission of HIV-1, indicate measures to reduce transmission by avoiding amniocentesis and breastfeeding and suggest that race and Rhesus blood type may be markers for genetic susceptibility to infection. (C) 1998 Rapid Science Ltd.