We compared with a new psychophysical method whether flashes and averted eye-gazes of a cartoon face induce a ventriloquist illusion (an illusory shift of the apparent location of a sound by a visual distracter). With standard psychophysical procedures that measure a direct ventriloquist effect and a ventriloquist aftereffect, we found in human subjects that both types of stimuli induced an illusory shift of sound location. These traditional methods, though, are probably contaminated by response strategies. We therefore developed a new two-alternative forced choice procedure that allows measuring the strength of an intersensory illusion in a bias-free way. With this new procedure we found that only flashes, but not averted eye-gazes, induced an illusory shift in sound location. This difference between flashes and eye-gazes was validated in an EEG study in which again only flashes illusorily shifted the apparent location of a sound thereby evoking a mismatch negativity response. These results are important because they highlight that commonly used measures of multisensory illusions are contaminated while there is an easy yet stringent way to measure the strength of an illusion in a bias-free way.