THE neotropical coral-snake complex with its contrasting ringed patterns includes true coral snakes, for example, the highly venomous elapid genus Micrurus, and many colubrids that are either less strongly venomous or non-venomous1. Whether or not this complex involves Batesian and/or Müllerian mimicry has been widely debated2–5. Wickler1 suggested that if mimicry depends on predator learning, true coral snakes are too deadly to be models; rather, they are mimics of the less dangerous colubrids of the complex (Mertensian mimicry). He also claimed that there was no evidence that any predator could recognise a coral snake innately1. I have shown that Costa Rican turquoise-browed motmots (Emomota superciliosa) need no learning to show strong aversion to a pattern of wide yellow and narrow red rings6; this I have interpreted as an innate recognition of a generalised coral-snake pattern. No Costa Rican member of the coral-snake complex has wider yellow than red rings, however, so I had no proof that the motmots were not simply showing an aversion to a general aposematic pattern. I present here evidence that another avian predator needs no learning to avoid not only a pattern of wide yellow and red rings, but also the most common local Micrurus pattern of red, yellow and black rings.