Laser Doppler anemometer (LDA) experiments were performed to gain quantitative information on the differences between the large-scale flow phenomena in a non-stenosed and a stenosed model of the carotid artery bifurcation. The influence of the presence of the stenosis was compared to the effect of flow pulse variation to evaluate the feasibility of early detection of stenosis in clinical practice. Three-dimensional Plexiglass models of a non-stenosed and a 25% stenosed carotid artery bifurcation were perfused with a Newtonian fluid. The flow conditions approximated physiological flow. The results of the velocity measurements in the non-stenosed model agreed with the results from previous hydrogen-bubble visualization. A shear layer separated the low-velocity area near the non-divider wall from the high-velocity area near the divider wall. In this shear layer, vortex formation occurred during the deceleration phase of the flow pulse. The instability of this shear layer dictated the flow disturbances. The influences of the mild stenosis, located at the non-divider wall, was mainly limited to the stability of the shear layer. No disturbances were found downstream of the stenosis near the non-divider wall. Using a pulse wave with an increased systolic deceleration time, the velocity distribution showed an extended region with reversed flow, a more pronounced shear layer and increased vortex strength. From these measurements it is obvious that the influence of the presence of a mild stenosis, mainly limited to the stability of the shear layer, can hardly be distinguished from the effects of a variation of the flow pulse. From this it can be concluded that methods for detection of mild stenosis, using solely the large-scale flow phenomena, as can be measured by ultrasound or MRI techniques, will hardly have any clinical relevance.