Defences against parasites are characterized by inducible, amplifiable responses, often with a memory component. Inducible defences with similar properties are common in a variety of other types of interactions, for example many aquatic invertebrates produce inducible structural defences against their predators and competitors. Most inducible defences have the following properties: (1) a threshold of activation; (2) an amplification of response with increasing stimulus; (3) a memory component. Specificity, amplification and memory are the basis for defining a defence as ‘immune’ (Klein, 1982), and these properties are present in both the vertebrate and invertebrate internal defence responses to pathogens. Invertebrates differ in the absence of immunoglobulins and therefore in reduced specificity. Although the reduced specificity of invertebrate internal defence systems is often viewed as proof of their ‘primitiveness’, the differences in defence systems of vertebrates and invertebrates may be more related to their respective selection regimes than to phylogeny. The syngeneic recognition system of vertebrates functions to recognize small departures from self, such as would arise from neoplasia. Are vertebrates under more intense selection from neoplasia, perhaps due to a greater incidence of hormonal imbalance or hypersensitivity reactions? The invertebrate internal defence systems are all less discriminating than the vertebrate, but there are marked differences in degree of discrimination depending on whether the group is colonial or not. Even the phyla of colonial animals with quite simple body plans, the sponges and cnidarians, have a more discriminating recognition system than the phyla of solitary animals with more complex body plans, such as the molluscs and arthropods. The primary effectors of all invertebrate responses to parasites are encapsulation and phagocytosis, although in some phyla there are specific antibacterial proteins that can also be induced. A primary effector of the vertebrate immune system is also based on phagocytosis, but is mediated largely by immunoglobulins. Although memory is a widespread property of inducible defence systems, the mechanisms, underlying memory components are unknown except in the vertebrate immune response. The function and adaptiveness of many of these accelerated or amplified second-set responses remain unclear, although it is tempting to conclude that a memory component is a beneficial adaptation to multiple, intermittent stimuli. From this review I conclude that the conditions that favour the evolution of inducible defences include: (1) an intermittent, unpredictable selective regime; (2) selection for a ‘camouflaged’ defence, to slow parasite counter-adaptation; (3) reliable cues and (4) high costs of defence.