In the illusory flash paradigm, a single flash may be experienced as two flashes when accompanied by two beeps or taps, and two flashes may be experienced as a single flash when accompanied by one beep or tap. The classic paradigm restricts responses to '1' and '2' (2-AFC), ignoring possible qualitative differences between real and illusory flashes and implicitly assuming that illusory flashes are indistinguishable from real flashes. We added a third response category 'different from that of either 1 or 2 flashes' (3-AFC). Eight naïve and 6 experienced observers responded to 160 real and 160 illusory flash trials. Experienced observers were exposed to 1,200 trials before the experiment but without receiving feedback on their performance. The third response category was used for only 4 % of the real flash trials and for 44 % of the illusory flash trials. Experienced observers did so more often (78 %) than naïve observers (18 %). This shows that observers can reliably identify illusory flashes and indicates that mere exposure to illusory flash trials (without feedback) is enough to detect and classify potential qualitative differences between real and illusory flashes.