While the first computer games synthesised all their sound effects, a desire for realism led to the widespread use of sample playback when technology matured enough to allow it. However, current research points to many advantages of procedural audio which is generated at run time from information on sound producing events using various synthesis techniques. A specific type of synthesis known as physical modelling has emerged, primarily from research into musical instruments, and this has provided audio synthesis with an intuitive link to a system’s virtual physical parameters. Various physical modelling techniques have been developed, each offering particular advantages, and some of these have been used to synthesise audio in interactive virtual environments. Refinements of these techniques have improved their efficiency by exploiting human audio perception. They have been implemented in large virtual environments and linked to third party physics engines, unveiling the potential for more realistic audio, reduced production costs, faster prototyping, and new gaming possibilities.