The risk of perinatal transmission of hepatitis C virus (HCV) from a cohort of 95 human immunodeficiency virus (HIV)-negative intravenous drug users (IVDU) is described, 89 of whom were positive for antibodies to HCV (anti-HCV). Infection, defined as the presence of HCV RNA in a serum sample collected from an infant at any time during follow-up, was detected in six of 63 (9.5%) infants born to HCV antibody-positive viraemic mothers. No mother who was HCV RNA negative at delivery transmitted HCV to her infant. Hepatitis C virus antibodies became undetectable in uninfected infants by 15 months, but persisted in all HCV-infected infants throughout follow-up. An abnormal alanine aminotransferase (ALT) level was observed on at least one occasion in all HCV-infected infants and in six occasions in uninfected infants. Two of the six HCV-infected infants became HCV RNA negative during follow-up by 27 and 29 months. Both of these infants had a large ALT elevation (mean peak ALT 398U l-1) at around 12 months of age. Analysis of a range of potential risk factors revealed that maternal HCV RNA load was important in predicting transmission, but suggested that other factors play a role in perinatal transmission from mother to child. No difference was found between mothers who transmitted HCV to their infants and those who did not for HCV genotype, duration of drug use, duration of methadone use, methadone dose, history of alcohol abuse, past hepatitis B virus (HBV) infection, mode of delivery, maternal and gestational age, birth weight and incidence of breast-feeding. Mothers who transmitted HCV to their infants had a longer duration between membrane rupture and delivery than the mothers who did not transmit (P = 0.03). HCV RNA was not detected in breast milk and colostrum samples from 38 viraemic mothers, including two who transmitted HCV to their infant.