Several studies have suggested that athletes with low hemoglobin saturation during exercise may experience impaired pulmonary blood gas exchange during maximal exercise. Blood viscosity may be implicated in exercise-induced pulmonary hemorrhage in race horses. We hypothesized that blood rheology may contribute to impaired gas exchange and reduced hemoglobin saturation during exercise in humans. A group of 20 highly trained endurance athletes participated in this study, 9 with low hemoglobin saturation during exercise (Low-SpO (2) group) and 11 with normal hemoglobin saturation (High-SpO (2) group). All subjects performed a progressive exercise test conducted to V.O (2max). Venous blood was sampled at rest, 50 % V.O (2max) and maximal exercise. Blood viscosity (etab) was measured at very high shear rate (1000 s (-1)) and 37 degrees C with a falling ball viscometer. The erythrocyte rigidity coefficient, "Tk", was calculated using the Dintenfass equation. At rest, no significant difference in etab was observed between the two groups (3.00 +/- 0.08 mPa . s vs. 3.01 +/- 0.04 mPa . s for the Low-SpO (2) and High-SpO (2) group, respectively). At 50 % V.O (2max) and maximal exercise, etab was higher in Low-SpO (2) (p < 0.01). Tk decreased in High-SpO (2) (p < 0.01) but remained unchanged in the other group during testing. The greater increase in etab in the Low-SpO (2) group during exercise may therefore have been due to the lack of reduction in Tk. As suggested by previous studies, the greater increase in blood viscosity in athletes with low hemoglobin saturation may lead to vascular shear stress. Whether this could impair the blood gas barrier and result in exercise-induced hypoxemia requires further study.