The effect of skin pigmentation on the reliability of noninvasive oximetry, especially during exercise and hypoxia, has not been thoroughly investigated. This is the first study, to our knowledge, that specifically addresses this question. Thirty-three young black men performed multistage, steady-state cycle ergometry, breathing gas mixtures simulating different altitudes: 33 breathed gas simulating sea level (PIO2 = 146 mm Hg), 11 breathed gas simulating 2,300 m (PIO2 = 110 mm Hg), and 22 breathed gas simulating 4,000 m (PIO2 = 85 mm Hg). Co-oximeter SaO2 determinations were performed in arterial blood samples obtained concurrently with ear oximetry that was measured using Hewlett-Packard 47201A (HP) and Blox IIA oximeters. The mean error or bias for the [HP - SaO2] and for [Biox IIA - SaO2] +/- 95% CI were: at simulated sea level (SaO2 greater than 96%): -0.4 +/- 0.3% and 2.1 +/- 0.3%; at simulated 2,300 m (range of SaO2 means, 89 to 94%): -0.8 +/- 0.5% and 3.5 +/- 0.9%; for simulated 4,000 m (range of SaO2 means, 75 to 84%): -4.8 +/- 1.6% and 9.8 +/- 1.8%, respectively. A better coefficient correlation was observed for all the pairs between SaO2 versus HP (r = 0.94, p less than 0.001, n = 279) than for the SaO2 versus Biox IIA (r = 0.80, p less than 0.001, n = 242). In conclusion, the HP oximeter appears to estimate SaO2 more accurately than the Biox IIA oximeter. The previously described overestimation for the Biox IIA ear oximeter and the underestimation for the HP ear oximeter at low SaO2 values in whites is exaggerated in blacks.(ABSTRACT TRUNCATED AT 250 WORDS)