The present paper shows the characteristics of a hardware/software display capable of presenting, at different spatial locations, multiple audio-visual stimuli in real-time. The system has been developed to provide a tool in neuroscience in event-related potential (ERP) studies, with presentation delays of less than one millisecond. The high time precision achieved makes it possible to overcome problems from previously used technologies in order to create a realistic environment of cross-modal stimulation. The design presented here has been built using a modular approach, which makes the system flexible and adaptable to different experiments or studies. The capability of presenting different stimuli either in audition, vision, or both combined in real-time, facilitates the study of realistic cross-modal attention experiments.