We studied 54 neonates with acute cardiorespiratory illness and 21 infants with bronchopulmonary dysplasia, to evaluate the accuracy of a nonheated pulse oximeter in predicting arterial oxygen saturation (SaO2). We also studied the accuracy of transcutaneous oxygen tension (tcPO2) in estimating arterial oxygen tension (PaO2) in infants with bronchopulmonary dysplasia. We compared pulse oximeter SaO2 with simultaneously measured SaO2 (range 78% to 100%) using a co-oximeter. Over a wide range of values for heart rate, blood pressure, hematocrit, PO2, PCO2, and pH, linear regression analysis revealed a close correlation between in vivo pulse oximeter readings and in vitro SaO2 measurements in patients with acute (r = 0.86, Y = 29.64 + 0.68X) and chronic (r = 0.91, Y = 6.29 + 0.96X) disease. Regression analysis of tcPO2 versus PaO2 showed an r value of 0.76 in infants with bronchopulmonary dysplasia. In these patients the mean difference between pulse oximeter SaO2 and in vitro SaO2 was 2.9% +/- 1.8% (SD), whereas the mean difference between tcPO2 and PaO2 was -14.5 +/- 11.1 mm Hg. Fetal hemoglobin ranged from 4.3% to 95%. We conclude that pulse oximetry is an appropriate alternative to tcPO2 for continuous oxygen monitoring in newborn infants with acute cardiorespiratory illnesses and chronic lung disease.