Animal communication is often deceptive; however, such dishonesty can become ineffective if it is used too often, is used out of context, or is too easy to detect [1, 2, 3]. Mimicry is a common form of deception, and most mimics gain the greatest fitness benefits when they are rare compared to their models [3, 4]. If mimics are encountered too frequently or if their model is absent, avoidance learning of noxious models is disrupted (Batesian mimicry [3]), or receivers become more vigilant and learn to avoid perilous mimics (aggressive mimicry [4]). Mimics can moderate this selective constraint by imperfectly resembling multiple models [5], through polymorphisms [6], or by opportunistically deploying mimetic signals [1, 7]. Here we uncover a novel mechanism to escape the constraints of deceptive signaling: phenotypic plasticity allows mimics to deceive targets using multiple guises. Using a combination of behavioral, cell histological, and molecular methods, we show that a coral reef fish, the dusky dottyback (Pseudochromis fuscus), flexibly adapts its body coloration to mimic differently colored reef fishes and in doing so gains multiple fitness benefits. We find that by matching the color of other reef fish, dottybacks increase their success of predation upon juvenile fish prey and are therefore able to deceive their victims by resembling multiple models. Furthermore, we demonstrate that changing color also increases habitat-associated crypsis that decreases the risk of being detected by predators. Hence, when mimics and models share common selective pressures, flexible imitation of models might inherently confer secondary benefits to mimics. Our results show that phenotypic plasticity can act as a mechanism to ease constraints that are typically associated with deception.