ABSTRACTThe world around us is full of beauty. Explaining a sense of the beautiful has beguiled philosophers and artists for millennia, but scientists have also pondered beauty, most notably Darwin, who used beauty to describe sexual ornaments that he argued were the subject of female mate choice. In doing so, he ascribed a ‘sense of the beautiful’ to non‐human animals. Darwin's ideas about mate choice and beauty were not widely accepted, however. Humans may experience beauty, but assuming the same about other animals risks anthropomorphism: we might find the tail of the peacock to be beautiful, but there is no reason to believe that peahens do. Moreover, mate choice, resurrected as an object of serious study in the 1970s, simply requires attraction, not necessarily beauty. However, recent advances in psychology and cognitive neuroscience are providing a new, mechanistic framework for beauty. Here we take these findings and apply them to evolutionary biology. First, we review progress in human empirical aesthetics to provide a biological definition of beauty. Central to this definition is the discovery that merely processing information can provide hedonic reward. As such, we propose that beauty is the pleasure of fluent information processing, independent of the function or consummatory reward provided by the stimulus. We develop this definition in the context of three key attributes of beauty (pleasure, interaction, and disinterestedness) and the psychological distinction between ‘wanting’ and ‘liking’. Second, we show how beauty provides a new, proximate approach for studying the evolution of sexual signalling that can help us resolve some key problems, such as how mating biases evolve. We also situate beauty within a more general framework for the evolution of animal signals, suggesting that beauty may apply not only to sexual ornaments, but also to traits as diverse as aposematic signals and camouflage. Third, we outline a variety of experimental approaches to test whether animal signals are beautiful to their intended receivers, including tests of fluency and hedonic impact using behavioural and neurological approaches.