This study is a retrospective analysis of the clinical data of 67 very low birth weight infants (VLBWI) with symptomatic patent ductus arteriosus (PDA); all were admitted to four neonatal intensive care units (NICU) from January 1, 1996 through December 31, 1996. The mean gestational age was 27.9 +/- 2.4 weeks, the mean birth weight was 1078 +/- 193 g. Fifty-six infants (83.6%) had RDS, and 53 infants (79%) received artificial surfactant. The NICU at a regional hospital used CVD score > or =3 as the sole criteria and the remaining three NICUs used color Doppler echocardiogram to confirm a symptomatic PDA and to treat it; sometimes LA/AO > or = 1.3 was used as the criteria for indomethacin treatment. Two of these three hospitals sometimes used the pulsed Doppler echocardiogram as well as color Doppler examination as the treatment criteria. Seven infants (10.5%) had contraindication for indomethacin treatment; four of them closed after conservative treatment, and another three were subjected to surgical ligation. Of the remaining 60 infants, 3 were treated with oral indomethacin and 57 were treated with intravenous indomethacin. The mean age when initial treatment given was 3.8 +/- 1.5 days (range, 8 hours approximately 20 days). Among them 10 (16.7%) were within 24 hours after birth, 25 (41.7%) were between 24 and 48 hours, and 25 (41.7%) were beyond 48 hours. The dosage of indomethacin was 0.2 mg/kg per dose intravenously every 12 to 24 hours for three doses as a full course, if not contraindicated. The mean dose of indomethacin was 2.8 +/- 1.5; 10 infants (16.7%) received 1 dose, 15 (25%) received 2 doses, 27 (45%) received 3 doses, 3 (5%) received 4 doses and 5 (8.3%) received 6 doses. Among them, 51 infants (85%) PDA closed (including 2 treated with oral indomethacin), 9 (15%) failed to close and 6 of them received surgical ligation (including 1 treated with oral indomethacin). The complications associated with indomethacin treatment were hypoglycemia (52%), decreased urine output (42%) and gastrointestinal hemorrhage (32%). The infants with RDS had an earlier mean age of initial treatment than non-RDS infants (3.3 +/- 2.5 vs. 7.6 +/- 5.6 days; p < 0.05), and also had a higher closure rate (89% vs. 57%; p < 0.05). There was a closure rate of 85% in this multicenter retrospective analysis. Even though the infants received only one or two doses, they still had a good chance of ductal closure (21/25, 84%). To minimize the complications associated with indomethacin treatment in VLBWI, the protocol of indomethacin treatment should be re-evaluated.