In the sound-induced flash illusion, discrepant numbers of stimuli in the visual and auditory modalities are integrated to create illusory visual percepts. While this phenomenon has been extensively studied over the last two decades, relatively little is known about the role of metacognition in this illusion. Specifically, when incongruent audiovisual signals yield the same behavioral reports as conditions with congruent audiovisual signals, is confidence in the perceptual judgment the same across conditions? To probe this question, we conducted an experiment where 22 observers viewed from 1-4 flashes and 1-4 beeps on each trial, and reported three things: (1) the number of perceived flashes, (2) confidence in the judgment about the number of flashes, and (3) confidence in whether the number of beeps and flashes were the same or different. In our exploratory analyses, we paired conditions of incongruent and congruent audiovisual signals which produced the same report about the number of flashes. Results showed that in several condition pairs, confidence in the perceived number of flashes was higher for congruent audiovisual signals than incongruent audiovisual signals. For confidence judgments about whether the number of auditory and visual signals presented were the same or different, confidence ratings were similar for all condition pairs but one. These findings provide preliminary support for the hypothesis that metacognition may be able to index differences between multisensory illusions and congruent multisensory information, but (in most circumstances) may be unable to index differences in the underlying causal structure which produces the sensory signals.