Changes in host behaviour following infection with parasites are frequently reported in the literature, and are often hypothesized to be adaptive for either host or parasite. However, investigators of such phenomena often use the "adaptation" label for host behavioural changes based on their intuition and not on rigorous criteria. Alterations in host behaviour following infection can only be considered adaptive if they satisfy certain conditions: (1) they must be complex; (2) they must show signs of a purposive design; (3) they are more likely to be adaptations if they have arisen independently in several lineages of hosts or parasites; and (4) they must be shown to increase the fitness of either the host or the parasite. A survey of published examples of host behavioural changes indicates that while some are spectacularly complex and are extremely well-fitted to their presumed function, most are simple increases or decreases in an activity performed prior to infection. There are some suggestions of convergent evolution in behavioural change in distantly related host or parasite groups but more evidence is needed. Finally, most known behavioural changes have not been demonstrated to lead to fitness gains in either hosts or parasites. Few known examples satisfy more than two of the above criteria, and, in general, the adaptive function of changes in host behaviour following infection is in need of more solid proof.