Natural selection should favour parasite genotypes that manipulate hosts in ways that enhance parasite fitness. However, it is also possible that the effects of infection are not adaptive. Here we experimentally examined the phenotypic effects of infection in a snail–trematode system. These trematodes (Atriophallophorus winterbourni) produce larval cysts within the snail's shell (Potamopyrgus antipodarum); hence the internal shell volume determines the total number of parasite cysts produced. Infected snails in the field tend to be larger than uninfected snails, suggesting the hypothesis that parasites manipulate host growth so as to increase the space available for trematode reproduction. To test the hypothesis, we exposed juvenile snails to trematode eggs. Snails were then left to grow for about one year in 800-l outdoor mesocosms. We found that uninfected males were smaller than uninfected females (sexual dimorphism). We also found that infection did not affect the shell dimensions of males. However, infected females were smaller than uninfected females. Hence, infection stunts the growth of females, and (contrary to the hypothesis) it results in a smaller internal volume for larval cysts. Finally, infected females resembled males in size and shape, suggesting the possibility that parasitic castration prevents the normal development of females. These results thus indicate that the parasite is not manipulating the growth of infected hosts so as to increase the number of larval cysts, although alternative adaptive explanations are possible.