It is widely accepted that 3D visualization of medical images helps in patient diagnosis. To visualize 3D images requires large computation loads, Therefore fast 3D visualization requires an expensive volume rendering board. Some web-based streaming techniques were suggested to construct high quality 3D visualizations even though only using a personal computer without a volume rendering board. These techniques could share the volume rendering board and diagnose 3D images together. To make a web-based 3D visualization system, the issues of time delay depending on limited network bandwidths need to be resolved, This delay directly influences the image quality, Through this study, we evaluated optimized streaming conditions to visualize and control the 3D medical image which has a high quality on the web. To construct this system, we primarily used three tools which included a VolumePro1000 board, WMV9 (Windows Media Video 9 codec) and TCP-IP based socket functions. The VolumePro1000 board could calculate quickly heavy matrixes of 3D images using Phong's shading and shear-warp factorization. WMV9 was able to efficiently compress live images and to apply image streaming techniques. TCP-IP based socket functions transmitted messages to control the 3D images. We developed a 3D visualization system and tested image qualities and transmission conditions of different compression rates for variable network conditions. It was very advantageous that the WMV9 encoding could be decoded automatically in many platforms (desktop, PDA, notebook, as long as the platform had Windows Media Player, We expect the 3D visualization system to be utilized in various biomedical fields such as IIGS (Interactive Image Guided Surgery), CAD (Computer Aided Diagnosis) and Tele-medicine technologies.