Ten healthy subjects performed two exercise sessions similar to an endurance training session for average athletes (1 cycling exercise at 85% of the maximal heart rate on a Monark cycle ergometer). In the first session, the subjects were not allowed to drink during exercise. During the second session, the same subjects performed the same exercise as during the first session but they drank a beverage volume equal to the weight loss induced by the first exercise session to verify the hypothesis that the compensation of sweat loss by drinking could attenuate the exercise-induced blood viscosity increase. Both protocols (with and without drinking) induced a significant increase of hematocrit and plasma viscosity. The whole blood viscosity increased at all shear rates but this increase was significant only for the exercise protocol without drinking. Blood thixotropic property, erythrocyte deformability, and erythrocyte aggregability remained unchanged after both exercise protocols. Hemoconcentration explained the increase of hematocrit and plasma viscosity. Hemoconcentration was probably the consequence of a filtration process through capillary leakage in addition to sweat loss, which could explain the partial effect of drinking in our study.