We compared automatic oscillometric measurements of systolic, diastolic, and mean arterial pressures with simultaneous pressures measured from arterial catheters in six preterm infants. Oscillometric arterial pressure measurements were performed with the Omega automatic instrument, using the recommended cuff and the size-larger cuff. Despite highly significant overall correlation coefficients for oscillometric and catheter measurements, 29% to 42% of oscillometric measurements with both cuff sizes were outside the 5-mm Hg error limits, and 4.0% to 9.3% were more than 10 mm Hg different from catheter measurements. Changing the cuff size did not reduce inherent errors in the oscillometric method.