We made an assessment of five pulse oximeters in regard to their ability to replace the HP ear oximeter as a noninvasive measurement of SaO2. Trials were performed during isocapnic progressive hypoxia (SaO2 range, 99 to 70 percent) in 22 white and six black subjects. Comparisons between values of SaO2 by oximetry were analyzed by comparing the difference of values by the two methods against their mean. The difference between pulse oximeters and the HP was 2.6 +/- 10.3 percent in all subjects. Results in whites were 1.9 +/- 10.2 percent and in blacks were 5.1 +/- 9.2 percent. The distribution of differences between pulse oximeters and the HP were larger below 80 percent than above 85 percent. We conclude that pulse oximeters give higher values than the HP, a tendency which is more pronounced in black than in white subjects. While the limits of agreement are better at saturations above 85 percent, the 95 percent confidence limits of agreement between pulse oximeters and the HP are rather large (+/- 10 percent) and unacceptable for assuming that pulse oximeters will provide the same values as found in clinical studies using the HP.