The ability to simulate realistic fluoroscopic images in real-time is a key aspect of any interventional radiology training system. In this paper, we propose a method for rendering X-ray images in real-time on a PC with consumer level graphics hardware, while improving the quality of the images. Although volume rendering techniques form the basis of our algorithm, we studied the characteristics of actual X-ray images to develop a method that can provide a new level of realism. In addition, the integration of the various levels of information contained in a CT scan in the rendering pipeline can be exploited to produce even more realistic, patient-specific X-ray or fluoroscopic images. Although the results presented here are preliminary, the performance of multi-texturing and multi-stage rasterization features available on recent low-cost graphics hardware already allows us to render X-ray images at about 30 frames per second.