In a randomized study, indomethacin was administered either by an intravenous rapid infusion or by a 30-minute intravenous infusion to preterm infants. Ten infants were randomly assigned to the rapid infusion group and nine to the other group. Time-averaged mean systolic, mean flow velocity, and mean diastolic blood flow velocities were measured in the internal carotid artery with a pulsed Doppler. Blood flow velocities, arterial blood gases, intraarterial blood pressures, and transcutaneous PO2 and PCO2 were measured at 5 and 1 minutes before indomethacin and then at 5, 10, 20, 30, 35, 40, 50, 60, and 90 minutes after indomethacin administration. Intravenous indomethacin either by rapid or slow infusion caused a significant decrease in the blood flow velocities. The blood flow velocities were significantly lower 5 minutes after rapid infusion and by 30 minutes in the slow-infusion group. In both groups, the blood flow velocities remained depressed up to 90 minutes. The blood pressures increased concomitantly to the blood flow velocities in the rapid-administration group only. Because indomethacin given for 30 minutes does not avoid the decrease in the blood flow velocities, further studies are necessary to ascertain the safest method of administration.