Although exogenous surfactants are of known efficacy in the prevention and treatment of respiratory distress syndrome (RDS), questions remain about the best regimens. During 1990-91, 6774 babies were recruited to an international multicentre trial to assess when administration of Exosurf, a synthetic surfactant, should be started and how often it should be given. The clinical outcome is known for 6757 (99.7%) infants. 2690 babies, judged to be at high risk of RDS when less than 2 hours of age, were randomly allocated to either early administration or delayed selective administration; 96% versus 73% received surfactant, at median ages of 118 and 182 min. The risk of death or dependence on extra oxygen at the expected date of delivery was 16% (95% CI 25% to 7%) lower among infants allocated early administration. Early administration was also associated with a 32% lower risk of pneumothorax. These 2690 infants were further randomised in a factorial design to either two doses of surfactant 12 hours apart, or the option of third and fourth doses at 12-36 hour intervals if signs of RDS persisted or recurred. 4067 other infants who later developed RDS were also recruited to this comparison, giving a total of 3376 infants allocated up-to-four doses (of whom, 45% received more than two) and 3381 allocated two doses. The outcome was similar in the two groups in respect of death, long-term oxygen dependence, and other major morbidity, even in secondary analyses restricted to infants who met the criteria for additional administration. There were more reports of poorly tolerated administration in the up-to-four doses group but no clear increase in serious morbidity, such as pulmonary haemorrhage. The OSIRIS trial suggests that early administration of surfactant to an estimated 32 babies, when compared with treatment of established RDS, would prevent 1 baby from dying and another from being dependent on extra oxygen long-term, but would entail the additional use of surfactant in 8 of these babies. It provides no evidence that a regimen including the option of third and fourth doses when signs of RDS persist or recur is clinically superior to a regimen of two doses.