Furosemide stimulates the renal synthesis of prostaglandin E2, a potent dilator of the ductus arteriosus. We administered this drug to 33 premature infants with the respiratory-distress syndrome, to determine whether it increased the incidence of patent ductus arteriosus. Chlorothiazide, a diuretic that does not stimulate prostaglandin E synthesis, was used as the control drug in 33 other infants. During the study, the incidence of patent ductus arteriosus was significantly higher (P less than 0.02) in the furosemide group (18 of 33 infants) than in the chlorothiazide group (8 of 33). Eleven infants in the furosemide group and seven in the chlorothiazide group required ductal ligation (P greater than 0.2). An additional six infants (all from the furosemide group) who did not have evidence of a patent ductus during the study were later found to have one. Overall survival was 76 and 61 per cent in the furosemide and chlorothiazide groups, respectively (P greater than 0.2). Small (less than twofold) increases in the urinary excretion of prostaglandin E were seen after the initial dose of both drugs. When the analysis was repeated after the fifth day of life, prostaglandin E excretion tripled after furosemide administration, whereas no increase occurred with chlorothiazide. We conclude that furosemide increases the incidence of patent ductus arteriosus in premature infants with the respiratory-distress syndrome, probably through a prostaglandin-mediated process.