Biological mimicry has long been viewed as a powerful example of natural selection's ability to drive phenotypic evolution, although continuing debates surround the mechanisms leading to its development and the nature of these mimetic relationships. Müllerian mimicry, in which unpalatable species derive a mutual selective benefit through evolved phenotypic similarity, has alternatively been proposed to evolve through either a two-step process initiated by a large mutational change, or through continuous gradual evolution toward a common aposematic phenotype. I exposed a model predatory fish species to two species of endemic Lake Tanganyikan Synodontis to provide evidence for aposematism and the presence of Müllerian mimicry in these species. Predators quickly became conditioned to avoid the venomous catfish and did not discriminate between the two species when they were switched, supporting a hypothesis of functional Müllerian mimicry in this group of similarly colored fish. Ancestral state reconstructions and statistical comparisons of color pattern divergence in Tanganyikan Synodontis indicate that Müllerian mimicry in these catfish has developed through diversification of an aposematic common ancestor with subsequent conservative mutualistic coevolution among its daughter lineages, rather than advergent evolution of a mimic toward a nonrelated model, as assumed by widely accepted models of Müllerian mimicry evolution.