Severe acute hypoxic respiratory failure is uncommon but often fatal. Standard treatment involves high inspired oxygen concentrations, mechanical ventilation and positive end-expiratory pressure. Many other interventions have been used in parallel with conventional treatment or as rescue therapy when it fails, including extracorporeal gas exchange, prone positioning, inhaled vasodilators, exogenous surfactants and drugs which modify the inflammatory process. Nearly all these treatments improve arterial oxygenation or markers of lung injury. However, the relationship between oxygenation and survival in acute hypoxaemic respiratory failure is not established, so improved oxygenation cannot be used as a surrogate for survival. Randomised controlled trials are, therefore, needed to assess the effects of these treatments on mortality. In such trials, extracorporeal oxygenation and extracorporeal carbon dioxide elimination, surfactant, early methylprednisolone, and prostaglandin E1 offer no survival advantage over conventional therapy. Prophylactic ketoconazole and pentoxifylline appear to improve mortality in small studies in surgical and oncology patients respectively, and methylprednisolone improves mortality and morbidity in unresolving disease.