Umbilical artery catheters are frequently used in the care of very low birth weight neonates to monitor arterial blood gas values. Historically, catheters with the tip placed low, in the abdominal aorta, have been associated with an increased risk of peripheral vascular complications. Recently, a report suggested that catheters placed high, in the thoracic aorta, were associated with an increased risk of intraventricular hemorrhage. To determine whether there is an association between the placement of an umbilical artery catheter high, in the thoracic aorta, and the occurrence of intraventricular hemorrhage (grades II through IV) or death within the first 5 days of life, a multicenter randomized trial was conducted. Nine hundred seventy neonates weighing 500 through 1499 g were randomly assigned to receive either high catheters, in the thoracic aorta closest to the sixth to eight thoracic vertebra, or low catheters, in the abdominal aorta closest to the third to fifth lumbar vertebra. The incidence of the primary outcome (intraventricular hemorrhage grades II through IV and/or death) was 25.8% among newborns with high catheters and 23.1% in those with low catheters (relative risk = 1.12, 95% confidence interval = 0.89, 1.39). However, a significant interaction was observed between the level of catheter placement and birth weight for neonates weighing 750 through 999 g and those weighing 1000 through 1499 g. The subgroup of neonates weighing 750 to 1000 g had a relative risk of the primary outcome of 0.72 (95% confidence interval = 0.49, 1.07), while those weighing 1000 through 1499 g had a relative risk of 2.02 (1.21, 2.36). The prevalence of peripheral vascular complications was 26.6% among neonates with low catheters compared with 16.0% among those with high catheters (relative risk = 1.66, 95% confidence interval = 1.30, 2.13). It is concluded that the risk for intraventricular hemorrhage (grades II through IV) and/or death among neonates weighing 500 through 1499 g with high catheters is not increased. Catheters placed low in the aorta, however, carry an increased risk of peripheral vascular complications. The observation that birth weight modifies the effect of catheter placement on the primary outcome must be cautiously interpreted because of the dangers of chance observations inherent in subgroup analyses.