Since pulse oximetry is now an ASA standard for intraoperative monitoring, we sought to determine the intraoperative failure rate for this device. We prospectively evaluated the intraoperative failure rate of our pulse oximeters at the four University of Washington Hospitals (University of Washington Medical Center, Veterans Affairs Medical Center [VAMC], Children's Hospital and Medical Center, and Harborview Medical Center [HMC]) recorded from April 1989 to August 1989. We defined failure as the inability to obtain any oximetry reading for a cumulative period of more than 30 minutes during any anesthetic procedure after all equipment malfunctions had been eliminated. Our puse oximeters failed in 124 of 11,046 cases studied; this is a failure rate of 1.12%, which ranged from 0.56% at HMC to 4.24% at VAMC. The failure rate at VAMC (4.24%) was significantly higher than the other hospitals (p less than 0.001). Those cases associated with the pulse oximeter failure had the following characteristics: (1) an ASA status of 3 or higher, (2) lengthy operations, and (3) elderly patients. When the device did fail in a patient, it did not function for 32% of the mean anesthesia time. We conclude that the intraoperative use of the pulse oximetry can provide information about blood oxygen saturation in most patients. However, in approximately 1% of the patients we studied in the operating room, mechanically functioning pulse oximeters failed to provide readings of blood oxygen saturations during routine operative use.