We examined the association between antenatal indomethacin exposure and adverse neonatal outcome in a matched retrospective cohort study of infants born to 72 mothers at less than 31 weeks' gestation. Indomethacin-exposed mothers were matched to controls by gestational age at delivery, antenatal corticosteroid exposure, prolonged spontaneous rupture of membranes, multiple pregnancy, thyrotrophin releasing hormone (TRH) exposure, and neonatal sex. Periventricular haemorrhage was significantly increased for infants delivered within 48 hours of maternal indomethacin exposure (Grade 1 and 2 19% versus 6%, and Grades 3 and 4 28% versus 3% (p<0.03)). Persistent patent ductus arteriosus was more common in those infants delivered within 48 hours of maternal indomethacin exposure (40% versus 20% (p<0.04)). More neonates exposed to antenatal indomethacin failed to respond to postnatal indomethacin to close a patent ductus arteriosus, 60% versus 0% (p<0.04). There were no adverse effects demonstrated of indomethacin administered greater than 48 hours from delivery. We have confirmed a probable association between antenatal indomethacin administration and an increased incidence of neonatal periventricular haemorrhage, patent ductus arteriosus, and impaired renal function. The adverse neonatal effects appear to be greatest when indomethacin is administered within 48 hours of delivery. We recommend that indomethacin should be used with caution as a tocolytic agent for the treatment of preterm labour at gestations less than 31 weeks.