DEFENSIVE mimicry has long been a paradigm of adaptive evolution by natural selection1-3. Mimics, models and predators in a batesian mimicry system (unpalatable model, palatable mimic) exist in a very different selective milieu from those in a müllerian system (involving >~2 unpalatable 'co-models')1,4-6. Consequently, the incorrect characterization of a mimicry relationship obscures the natural histories of populations involved and undermines attempts to test general mimicry theory by means of empirical studies of specific systems. Here, we reassess the classic case of mimicry involving viceroy butterflies, Limenitis archippus (Cramer) (Nymphalidae), and two species they purportedly mimic: the monarch, Danaus plexippus (L.), and the queen, Danaus gilippus (Cramer) (Nymphalidae: Danainae). Viceroys are historically considered palatable (batesian) mimics7,8 of the chemically defended9 danaines. Our experiment refutes this interpretation by revealing that viceroys are as unpalatable as monarchs, and significantly more unpalatable than queens from representative Florida populations. This implies that viceroys are müllerian co-mimics of the danaines and prompts a comprehensive reassessment of this widely cited exemplar of mimicry.