It is now well recognized that blood is a yield stress fluid, that is to say that it will not flow below a critical threshold level of stress. This is reflected in vivo with a cessation of flow threshold, of great importance in circulatory physiology and pathophysiology. Values for yield stress obtained by different techniques are scarce. Rheometry at low shear rates is one method of obtaining values. After mitigating the slip effects on the walls of the rheometer, it has been possible to get an approximate value of the yield stress in human blood samples in controls and in ill subjects with different hematocrits. A Low Shear 40 rheometer with a controlled rotation speed was used. Home made geometries were used and the shear rate was varied in the range 10(-3) s-1-100 s-1. Fresh blood samples from healthy donors and patients were anticoagulated with EDTA. Suspensions of red cells in plasma were prepared by removing plasma. The measurements were made at 25 degrees C within four hours following blood sampling. Particular attention was paid to data acquisition at low shear rates (10(-3) s-1, 3.10(-3) s-1, 10(-2) s-1, 3.10(-2) s-1) with a computer linked to the rheometer. Shear stress approached a constant value at low shear rates for geometries with rough surfaces. Shear stress measured at 10(-3) s-1 was taken as an approximation of yield stress. Values were 5 to 6 mPa for hematocrit 60% and 20 to 25 mPa for hematocrit in the 80 to 90% range. For pathological blood samples, accurate measurements were made at shear rate as low as 0.01 s-1. Migrational effects were present at 10(-1) s-1 et 3.10(-3) s-1 but they could be greatly mitigated by the use of the 170 microns surface roughness.