The stress in the endothelial cells induced by blood flow depends on the waviness of the blood-endothelium interface and the slopes at the junctions of neighboring cells in the direction of flow. The height and slope in the third dimension of the living endothelial cells cannot be measured by ordinary optical and electron microscopy. Here we show that interference microscopy meets the challenge. We measured the geometry of cultured confluent human vascular endothelial cells in a flow, and we found that in a normal section parallel to the flow, the absolute values of the surface slopes at the cell junctions were 0.70 ± 0.02 (SE) and 0.80 ± 0.02 (SE) at the leading and trailing edges of the cells, respectively, in a culture medium of osmolarity 310 mosM with a shear stress of approximately 1 N/M2. A reversal of the flow direction led to a reversal ofthe slope pattern. An increase in medium osmolarity above 310 mosM induced an initial decrease in the slopes followed by a return to normal, whereas a decrease in the osmolarity had a reversed effect. These results, in light of our previous theoretical analyses, show that tensile stress exists in the endothelial cell membrane, and that the mechanism of tension accumulation is a reality. The accumulation is not 100% because the membranes are not smooth at the cell junctions.the membrane of cell i + 2 must bear not only the shear acting on it directly but also the cumulative tension in the membrane of cell i + 1. In either case the mechanism of accumulation can become serious because hundreds of thousands of cells are on line. The reality is closer to Fig. 1A. Ifthe in vivo blood shear is 1 N/M2, then the tension in the cell membrane can be 104 N/M2 or larger, which causes a shear stress of 5 x 103 N/M2 or larger acting on planes inclined at 450 to the direction of tension. If integrins or ion channels were lined up with these inclined planes, they could, theoretically, be sensitive to the large shear stress. Thus the significance is seen.In the past, the significance of the membrane geometry of the endothelial cells was unknown, and there is no known method for its determination. An atomic force microscope may be applied (8), but it is difficult to use it in a flowing fluid. Electron microscopy requires tissue fixation and dehydration, which can induce geometric distortion. Facing these difficulties, we recalled our early work on the determination of the thickness profile of erythrocytes by using a MachZender interference microscope (9-13), which yielded data with a resolution of 0.02 ,um based on physical optics. Hence we hypothesized that it might yield the endothelium profile. This paper is concerned with the geometric shape of the interface between flowing blood and a blood vessel. The endothelium, which is a confluent layer of endothelial cells, is the biological gateway between the blood and the tissues and organs of the body and is the source of many factors that are critical to health and disease such as atherosclerosis. As sketched in Fig. 1, each endot...