We investigate evolution of two categories of adaptive host manipulation by trophically transmitted helminths: (1) predation suppression decreases the host's mortality before the helminth is capable of establishing in its next host; (2) predation enhancement increases the existing host's mortality after it can establish in its next host. If all parasite mortality is purely random (time-independent), enhancement must increase predation by the next host sufficiently more (depending on manipulative costs) than it increases the average for all forms of host mortality; thus if host and parasite die only through random predation, manipulation must increase the “right” predation more than the “wrong” predation. But if almost all parasites die in their intermediate host through reaching the end of a fixed life span, enhancement can evolve if it increases the right predation, regardless of how much it attracts wrong predators. Although enhancement is always most favorable when it targets the right host, suppression aids survival to the time when establishment in the next host is possible: it is most favorable if it reduces all aspects of host (and hence parasite) mortality. If constrained to have selective effects, suppression should reduce the commonest form of mortality.