In this paper, we present a new approach to computer-aided training in Medicine based on the visualisation of simulated clinical conditions using a virtual patient. Our system uses a knowledge-based model of cardiac physiology to generate realistic emergency situations (shock states). The overall results of the simulations are presented through a realistic 3D virtual patient whose behaviour and appearance are consistent with the clinical data generated. After introducing the principles behind the simulation and the system?s architecture, we describe the process through which physiological parameters are mapped to the visual appearance of the 3D patient and discuss representative results from the system.