Abstract: With the continuous development of micro-systems, micro-control and relevant technologies, the demands for micro-measurements in industrial automations of production and inspection are increasing rapidly. Meanwhile, more critical requirements accuracy, efficiency, and cost are expected. To address this situation, a digital micro-scopic depth measurement system is constructed with a common stereo microscope and two cameras. In the work, a specific fixture is designed to connect the cameras and the microscope to enable easy optical alignment and calibration. Then a formula is summarized to build an accurate relation between disparity and depth corre-sponding to the specific microscopic system. Next, a CUDA enhanced fast stereo matching algorithm is proposed and practiced to give live disparity map. Finally, the disparity map is transformed into the depth map so that can achieve the depth measurement of microscopic target and perform accuracy analysis. The adjustable magnifi-cation of microscope gives feasibility for different view cope and precision requirement. The system takes several types of microcapacitors as the object of analysis, and quantitatively studies the relevance between measure-ment resolution-parallax-precision at the same time. Compared with some existing research, experimental results show a micron level precision and great efficiency that promises various real applications.