The reliability and accuracy of an intra-arterial fluorescent optode system to measure continuously pHa, PaCO2, and PaO2 were evaluated in a dog model and in a group of critically ill patients. Six hundred sixty-three arterial blood gas (ABG) samples were analyzed for pHa, PaCO2, and PaO2 in the dog studies. The intra-arterial blood gas system (IBGS) indicated a steady state in 420 instances for pH, 359 instances for PaCO2, and 256 instances for PaO2. Comparison of these ABG and IBGS values by linear regression analysis revealed r = .97 for pHa, .95 for PaCO2, and .96 for PaO2. The mean +/- SD of the difference between ABG and IBGS was -0.02 +/- 0.03 for pHa, 1.05 +/- 3.8 for PaCO2, and -17 +/- 13 for PaO2. Nonsteady states were correctly identified by the IBGS in every instance. Comparisons between 79 temporally matched ABG and IBGS values, exclusive of in vivo calibration samples, in 12 critically ill patients revealed r = .97 for pHa, .96 for PaCO2, and .99 for PaO2. The difference was 0.002 +/- 0.02 for pHa, 0.44 +/- 2.97 for PaCO2, and -1.22 +/- 9.34 for PaO2. We conclude that it is possible to measure continuously pHa, PaCO2, and PaO2 with the IBGS in critically ill patients for periods from 3 to 25.5 h while maintaining the ability to monitor BP and withdraw blood samples from the arterial cannula. Agreement between the two techniques is within clinically acceptable ranges for pHa and PaCO2, whereas PaO2 measurement by the IBGS requires further refinement.(ABSTRACT TRUNCATED AT 250 WORDS)