We discuss the problems of spatial perception that arise by the transmission of endoscopic surgery by traditional video systems and analyse the improvements that can be achieved by two- und three-dimensional systems. Starting with a description of the physics of vision and the principles of two- and three-dimensional imaging, we demonstrate the importance of optimal adaptation of technology to the natural perception processes of the observer. In this context we review critically the technology that is presently available. Only stereoscopic video technology with 'shutter glasses' provides the observer with the spatial information that is of such decisive importance for minimally invasive surgery. Finally we discuss new technical developments that further improve three-dimensional vision in endoscopy, such as a 3D video module that is small enough to be used intracorporally.