BACKGROUND Patent ductus arteriosus is a common complication of prematurity that frequently requires surgical or medical treatment. The benefit of prophylactic treatment by indometacin, a cyclo-oxygenase inhibitor, remains uncertain compared with curative treatment. This benefit could be improved with ibuprofen, another cyclo-oxygenase inhibitor with fewer adverse effects than indometacin on renal, mesenteric, and cerebral perfusion. We aimed to compare prophylactic and curative ibuprofen in the treatment of this abnormality in very premature infants. METHODS We did a randomised controlled trial in infants younger than 28 weeks of gestation, who were randomly assigned to receive either three doses of ibuprofen or placebo within 6 h of birth. After day 3, symptomatic patent ductus arteriosus was treated first by open curative ibuprofen, then back-up indometacin, surgery, or both. The primary endpoint was need for surgical ligation. Analysis was per protocol. RESULTS The study was stopped prematurely after 135 enrollments because of three cases of severe pulmonary hypertension in the prophylactic group. 65 infants received prophylactic ibuprofen, and 66 received placebo. Prophylaxis reduced the need for surgical ligation from six (9%) to zero (p=0.03), and decreased the rate of severe intraventricular haemorrhage from 15 (23%) to seven (11%) (p=0.10). However, survival was not improved (47 [71%] placebo vs 47 [72%] treatment, p=1.00), because of high frequency of adverse respiratory, renal, and digestive events. CONCLUSIONS In premature infants, prophylactic ibuprofen reduces the need for surgical ligation of patent ductus arteriosus, but does not reduce mortality or morbidity. Therefore, it should not be preferred to early curative ibuprofen.