Acute upper-gastrointestinal bleeding is a common indication for emergency admission to hospital. Risk-stratification scores have been devised to identify patients at risk of re-bleeding or death, but these have usually required both clinical and endoscopic assessment. Two recent studies have employed clinical criteria alone to identify low-risk patients that may not require admission or in-patient endoscopy. While each of these studies has individual merit, both are unable to answer the question of whether risk stratification improves health outcomes or resource use in acute gastrointestinal haemorrhage. They will nevertheless help the development of guidelines that enable patients to be managed more efficiently and outcomes to be compared more fairly.