How a parasite (or its offspring) moves from onehost to the next remains a central topic in parasitol-ogy. Understanding such strategies is at the heart ofapplied aspects of parasitology, but it is also im-portant for solving more basic biological questions.One strategy of transmission that is especially in-triguing is that of host manipulation, which occurswhen a parasite enhances its own transmission byaltering host behaviour. We begin this paper witha brief historical overview of the ‘manipulation hy-pothesis,’ in order to illuminate past and present re-search on this transmission strategy, as well as currentchallenges.Scientists were beginning to suspect that parasitescould manipulate their hosts early in the 20th century(e.g. Cram, 1931). In 1952, van Dobben reported thatfish retrieved from cormorants (definitive hosts) werefar more likely to be intermediate hosts of the cestodeLigula intestinalis than were fish captured by fisher-men. Bethel and Holmes (1973, 1977) used labora-tory experiments to show that the cystacanths of theacanthocephalan Polymorphus paradoxus provoke ab-normal behaviours in the amphipod (Gammarus la-custris ; intermediate host), and then verified the re-sulting increased predation risk from ducks (definitivehosts).Since that time, there has been increasing enthusi-asm among parasitologists for the study of phenotypicchanges in parasitised animals. The idea that parasitescould manipulate the phenotype of their host and thusenhance their own transmission became rapidly popu-larnotonlybecauseitwasinherentlyafascinatingphe-nomenon, but also because it offered parasitologists anopportunity to demonstrate the ubiquitous importanceof parasites to a broader community of scientists. Dueto an impressive number of studies performed duringthelastthreedecadesonthistopic,parasite-inducedal-terations of host phenotypes are now documented for awiderangeofparasites(seeBarnardandBehnke,1990;Combes, 1991, 1998; Poulin, 1998; Moore, 2002 forreviews). These studies have demonstrated that a largerange of host phenotypic traits can be altered by para-sites (e.g. behaviour, morphology and/or physiology),and that the alterations can vary greatly in their magni-tude, from slight shifts in the percentage of time spentinperformingagivenactivitytotheproductionofcom-plex and spectacular behaviours (Poulin and Thomas,1999; Moore, 2002).The most popular example of parasitic manipula-tion in ecological textbooks seems to be the trema-tode“brainworm” Dicrocoeliumdendriticum .Ants(in-termediate hosts) infected with this trematode ascendblades of grass, a behaviour that probably enhancestransmission to grazing sheep. However, this is not thebest exemple of parasitic manipulation. As one mightimagine, it is difficult to study ant predation by sheep,and the relative numbers of infected and uninfectedants that are eaten by these herbivores remain a mys-tery. There are however many impressive examples ofapparent host manipulation that are more amenableto quantification. For instance, numerous trophicallytransmitted parasites have been shown to alter the be-haviour of their intermediate hosts in a way that in-creases their vulnerability to predatory definitive hosts(Lafferty,1999;Berdoyetal.,2000;Moore,2002).Par-asites also manipulate host habitat choice; arthropodsharbouring mature nematomorphs or mermithids seekwater and jump into it, thereby allowing the parasiticworm to reach the aquatic environment needed for itsreproduction (Thomas et al., 2002a). Mermithid nema-todes can also feminize male insect behaviour whenparasitetransmissionisdependentonafemale-specificbehaviour (Vance, 1996). Parasitic wasps can maketheir spider host weave a special cocoon-like structuretoprotectthewasppupaeagainstheavyrain(Eberhard,2000, see also Brodeur and Vet, 1994), or can evencause the host to seek protection within curled leavesto protect pupae from hyperparasitoids (Brodeur andMcNeil, 1989). Viruses may stimulate superparasitismbehaviour in solitary parasitoids and thus achieve hor-izontal transmission (Varaldi et al., 2003). Some dige-neans drive their molluscan intermediate hosts towardideal sites for the release of cercariae (Curtis, 1987).‘Enslaver’ fungi make their insect hosts die perchedin a position that favors the dispersal of spores by thewind (Maitland, 1994). Vector-borne parasites can ren-der their vertebrate hosts more attractive to vectors,and/orcanmanipulatethefeedingbehaviourofvectorsto enhance transmission (Hamilton and Hurd, 2002).Allthesespectacularphenotypicchangeshavebeenin-terpreted as the sophisticated products of natural selec-tionthathasfavoredhostmanipulation,thusincreasingthe likelihood that parasite propagules will encounterthe next host or a suitable habitat. From an evolution-ary point of view, these changes are classically seen ascompelling illustrations of the ‘extended phenotype’