In the Sound-Induced Flash Illusion (SIFI) sound dramatically alters visual perception, as presenting a single flash with two beeps results in the perception of two flashes. In this comprehensive review, we synthesise 20 years of research using the SIFI, from over 100 studies. We discuss the neural and computational principles governing this illusion and examine the influence of perceptual experience, development, ageing and clinical conditions. Convergent findings show that the SIFI results from optimal integration and probabilistic inference and directly reflects crossmodal interactions in the temporal domain. Its neural basis lies in early modulation of visual cortex by auditory and multisensory regions. The SIFI shows increasingly strong potential as an efficient tool for measuring multisensory processing. Greater harmonisation across studies is now required to maximise this potential. We therefore propose considerations for researchers relating to choice of stimulus parameters and signpost directions for future research.