We developed an accurate three-dimensional blood vessel search (3D BVS) system using NIR light for the clinical blood sampling. In the previous study, the 3D BVS system, which used near-infrared (NIR) light imaging and the stereo method to locate blood vessel accurately in three dimensions has been developed(1). However, as NIR lights could not transmit the human arm, this system could not be used for the subcutaneous blood vessel detection. In this study, we developed a BVS by using the reflecting NIR light for blood sampling assist. The light scattering in human tissue will cause blur of blood vessel edge in image, that makes the diameter of blood vessel became uncertain. In this study, a light propagation simulation and a multilayer phantom were adopted to estimate the measurement error of blood vessel diameter in our BSV system. In the simulation, the optical properties of scattering coefficient, absorption coefficient, and refractive index were set similar with human skin. Next, we fabricated a multilayer phantom, which has the similar structure and optical properties with the human skin to confirm availability of the simulation. Also, the optical properties of our phantom are adjustable in our phantom to imitate the different color of skin. We established the estimation algorithm to detect the blood vessel accurately. Finally, we confirm the availability of our BVS for the blood sampling assist system.