Blood viscosity depends essentially on two factors: shearing rate, since the blood is not a Newtonian liquid and the hematocrit, any fall in which tends to cause the blood to become closer in its behaviour to a Newtonian liquid. Furthermore, these two factors are interdependent. When the hematocrit falls, blood viscosity decreases in greater proportions at lower shearing rates than at high shearing rates. Whilst in vitro, in hemodilution, results vary in relation to the type of replacement solution, in vivo this factor would appear to be of little importance by virtue of the adapted variations in the deformability of the red cells. However, hemodilution results in the greatest relative decreases in blood viscosity at the highest hematocrits. Below a hematocrit of approximately 30 p. cent, the gain in terms of viscosity is very low. This figure would appear to be the ideal hematocrit level from a rheological standpoint. An additional rheological advantage is provided by hemodilution the instability of red cell rouleaux, which are in fact created more readily in areas where shearing rate is low. All of these rheological changes explain the hemodynamic consequences.