In animal communication, an alarm signal is an antipredator adaptation in the form of signals emitted by social animals in response to danger. Many primates and birds have elaborate alarm calls for warning conspecifics of approaching predators. For example, the alarm call of the blackbird is a familiar sound in many gardens. Other animals, like fish and insects, may use non-auditory signals, such as chemical messages. Visual signs such as the white tail flashes of many deer have been suggested as alarm signals; they are less likely to be received by conspecifics, so have tended to be treated as a signal to the predator instead.

Different calls may be used for predators on the ground or from the air. Often, the animals can tell which member of the group is making the call, so that they can disregard those of little reliability.

Evidently, alarm signals promote survival by allowing the receivers of the alarm to escape from the source of peril; this can evolve by kin selection, assuming the receivers are related to the signaller. However, alarm calls can increase individual fitness, for example by informing the predator it has been detected.

Alarm calls are often high-frequency sounds because these sounds are harder to localize.

This cost/benefit tradeoff of alarm calling behaviour has sparked many interest debates among evolutionary biologists seeking to explain the occurrence of such apparently "self-sacrificing" behaviour. The central question is this: "If the ultimate purpose of any animal behaviour is to maximize the chances that an organism's own genes are passed on, with maximum fruitfulness, to future generations, why would an individual deliberately risk destroying itself (their entire genome) for the sake of saving others (other genomes)?".

Some scientists have used the evidence of alarm-calling behaviour to challenge the theory that "evolution works only/primarily at the level of the gene and of the gene's 'interest' in passing itself along to future generations." If alarm-calling is truly an example of altruism, then human understanding of natural selection becomes more complicated than simply "survival of the fittest gene".

Other researchers, generally those who support the selfish gene theory, question the authenticity of this "altruistic" behaviour. For instance, it has been observed that vervets sometimes emit calls in the presence of a predator, and sometimes do not. Studies show that these vervets may call more often when they are surrounded by their own offspring and by other relatives who share many of their genes. Other researchers have shown that some forms of alarm calling, for example, "aerial predator whistles" produced by Belding's ground squirrels, do not increase the chances that a caller will get eaten by a predator; the alarm call is advantageous to both caller and recipient by frightening and warding off the predator.

Another theory suggests that alarm signals function to attract further predators, which fight over the prey organism, giving it a better chance of escape. Others still suggest they are a deterrent to predators, communicating the prey's alertness to the predator. One such case is the western swamphen (Porphyrio porphyrio), which gives conspicuous visual tail flicks (see also aposematism, handicap principle and stotting).