AIM: To assess the role of etamsylate in reducing the risk of haemorrhagic brain damage and its consequences. DESIGN: Follow up of babies recruited into a randomised controlled trial. METHODS: A total of 334 infants born before 33 weeks gestation in France and Greece were randomly allocated within the first four hours of birth either to receive etamsylate or to act as controls. The principal outcomes in the trial were death or impairment and/or disability at the age of 2 years. RESULTS: Fifty nine children were lost to follow up. A total of 115 (34%) either died or had some impairment or disability, and 88 (26%) either died or had severe impairment or disability at 2 years of age. These outcomes did not differ significantly between the two randomised groups: relative risks and 95% confidence intervals 1.14 (0.78 to 1.4) and 1.17 (0.82 to 1.68) respectively. The findings were similar for all the prespecified subgroup analyses stratified by key prognostic factors at trial entry: country of birth, gestational age < or >or= 29 weeks, inborn or outborn, age < or >or= 1 hour, and with or without cerebral scan abnormality. CONCLUSION: These findings do not support the use of etamsylate. Other strategies need to be evaluated for the prevention of mortality and morbidity in these vulnerable infants.