ObjectiveThe aim of the study was to test the reliability and variation in the readings of two widely used pulse oximeters in preterm infants.DesignTwo different pulse oximeters and a transcutaneous Po2monitor were used to record the data continuously on a cotside computer database.PatientsSixteen preterm infants were studied in the Neonatal Unit, Simpson Memorial Maternity Pavilion, Edinburgh, UK.InterventionsNone.Measurements and Main ResultsApproximately one fifth of the time, the pulse oximeter readings could be established as artifactual. Study of the remaining four fifths of the data showed that, on average, the Nellcor pulse oximeter recorded saturation percentages 2.2% higher than the Ohmeda oximeter.ConclusionsWe recommend that all neonatal units adopt a policy of using different saturation alarm limits for these two instruments. We further recommend that other pulse oximeters be tested by a methodology similar to the one we present in this paper, before their use in monitoring oxygenation in preterm infants.(Crit Care Med 1997; 25:2051-2054)