A system is introduced, which can automatically measure and analyze a cephalogram. This system includes the patient information management module, the landmarking module, the measuring and analysis module, and the inter-active module. Several image processing methods are used in the course of landmarking, such as wavelet multi-scale analysis and Canny filter. A knowledge-based edge-tracing algorithm also is used to detect edges. Landmarks are detected according to the facial anatomy knowledge. The experimental result shows that the system has good performance.